The previous two articles in this series catalogued the scale of IT sector layoffs globally and Anthropic’s specific role in accelerating the employment disruption. This article is different in purpose and in audience. It is written for a single, specific reader: a US citizen who works in engineering or technology, who is watching the headlines about 30,000 Amazon jobs and 15,000 Microsoft jobs and wondering what it means for them, and who wants evidence-based guidance rather than platitudes about “embracing change.”
The US citizen framing is deliberate and important. Not because non-citizens do not face the same pressures - they face more, with immigration timelines layered on top of employment uncertainty - but because the citizen status unlocks a set of career pathways, job categories, and strategic options that are entirely unavailable to engineers on H-1B visas, green card holders in priority queues, or international professionals working at US subsidiaries. Clearance-eligible work, certain government contracting roles, specific defence technology positions, and federal employment itself are citizenship-gated to varying degrees. In a market where traditional Big Tech software engineering roles are contracting and where understanding which growing sectors require citizenship is a strategic edge, being deliberate about that advantage matters.
This article covers six interconnected areas. The first is an honest assessment of which roles are dying and at what pace, without euphemism. The second is a detailed breakdown of which roles are genuinely growing, with verified salary data. The third is a geographic analysis of where to be and where not to be in the next two years. The fourth is a set of citizenship-specific career pathways that are either exclusively or preferentially available to US citizens. The fifth is a skills and retraining roadmap with specific timelines, resources, and ROI estimates. The sixth is a 90-day action plan for engineers at each of three career stages: early career (zero to five years), mid-career (five to fifteen years), and senior (fifteen or more years).
No article of this type can promise specific outcomes in a market this volatile. What it can provide is the most accurate available map of the territory, based on verified salary data, hiring trend analysis, and the trajectory of AI capability development as documented by Anthropic’s own Economic Index and the broader evidence base assembled in this series. The decisions belong to each reader. The goal of this article is to make those decisions as well-informed as possible.
Part One: The Honest Assessment - What Is Actually Being Eliminated
Before mapping the opportunities, it is necessary to be precise about the threat. The layoff wave of 2025 and 2026 is being discussed in terms so abstract that many engineers who are in the highest-risk categories are not recognising their own situation. This section names the specific functions being eliminated and the mechanisms by which they are being eliminated, so that anyone in those functions can calibrate their urgency accurately.
Manual Software Testing and Quality Assurance
The median QA engineer salary in the United States is approximately $95,000 to $115,000 depending on location and experience level. This role is the single most directly and comprehensively automated function in the software development lifecycle. AI testing tools including Selenium with AI extensions, Mabl, Testim, and GitHub Copilot’s test generation capabilities can now generate, execute, and report on test suites at a scale and speed that no manual QA team can match.
The documentation of this displacement is specific. Multiple major tech companies have confirmed reductions of 20% to 40% in manual QA headcount over the last 18 months, with the productivity gap between AI-assisted and manual testing widening with each quarterly model update. The BLS projects the occupation of “Software Quality Assurance Analyst” to grow more slowly than average through 2033. Anthropic’s Economic Index shows high observed AI coverage of the tasks that constitute the QA role.
The engineers in this category who are most at risk are those whose entire skill portfolio is manual testing: test case writing, bug documentation, regression execution, and UAT coordination. Those who have added automation skills - programming AI testing frameworks, designing test architectures, managing AI-powered test platforms - are in a genuinely different position, but the distinction requires active skill investment to maintain.
For any US engineer currently working primarily in manual QA, the signal from the data is urgent: the three-year outlook for roles built entirely around manual test execution is poor, and the transition to test automation or test architecture needs to begin now, not when the next round of layoffs is announced.
Routine Software Development - The Entry Level Squeeze
Entry-level software engineering hiring has declined approximately 25% at the fifteen largest tech companies between 2023 and 2024, according to SignalFire data cited in IEEE Spectrum. That trend has continued and in some categories accelerated through 2025 and into 2026, as Claude Code, GitHub Copilot, and similar tools allow mid-level and senior engineers to produce significantly higher volumes of code per hour.
The functions most directly displaced at the entry level are: boilerplate code generation, CRUD application development, routine API integration, bug fixing in well-understood codebases, and documentation generation. These were the primary tasks that entry-level engineers built their initial career capital around - the work that gave them practical experience with real systems. When that work is absorbed by AI tools, the entry-level pipeline that historically fed the senior engineering roles of the next decade is structurally compressed.
This creates two simultaneous problems. The immediate problem is that recent CS graduates and engineers with one to three years of experience are competing for fewer positions at entry level. The structural problem is that the traditional apprenticeship model - where junior engineers learned by doing the routine work that seniors delegated - is being disrupted at its foundation. If AI does the routine work, the pathway to becoming a non-routine senior engineer becomes less clear.
Engineers in this position have specific options that are developed in the retraining section of this article. The key point here is that optimising for traditional junior software engineering positions at large tech companies is a lower-return strategy in 2026 than it was in 2022, and engineers who recognise this shift earlier have more options for the alternative paths.
Project and Programme Management in Tech
Programme management - the coordination and administration layer that has been one of the most reliable career progression paths for mid-career tech workers - is the second major category of direct AI displacement. Tools including Microsoft Copilot, Claude with project management integrations, AI-assisted Jira and Asana management, and automated meeting summary and action item extraction have absorbed significant portions of the work that programme managers were hired to perform.
The Microsoft July 2025 cuts specifically targeted programme management layers, with the company acknowledging that Copilot’s capabilities in status reporting, cross-team dependency tracking, and stakeholder communication had reduced the human hours required in these functions. Amazon’s “anti-bureaucracy” framing for its 30,000 cuts was primarily an attack on the programme management and coordination layers between technical teams.
The PM roles most at risk are those built around coordination, documentation, and status management - the “human middleware” functions that AI can now partially replace. The PM roles most resilient are those involving genuine strategic judgement, client or executive relationship management, complex technical trade-off decisions, and the political and interpersonal navigation of large organisations that AI cannot replicate.
For engineers who have moved into programme management as a career progression, the question is whether their current role is primarily coordination work (high AI exposure) or primarily judgement and relationship work (lower AI exposure). The distinction determines both the urgency of the situation and the appropriate strategic response.
Technical Writing and Documentation
Technical writing - producing API documentation, user guides, internal system documentation, and developer tutorials - is being rapidly automated by AI tools. The observation in Anthropic’s Economic Index that Technical Writers show high AI task coverage reflects the straightforward nature of the displacement: documentation tasks involve synthesising known information into structured formats, a workflow that AI handles effectively.
The US Bureau of Labor Statistics projects technical writing employment to decline through the forecast period. This is a role where the traditional advice to “learn AI tools” is genuinely insufficient, because the tools are not merely making technical writers more efficient - they are substituting for the core function. A company that could previously justify three technical writers can often now produce equivalent documentation with one writer using AI tools, if their documentation needs are for standard API and software products.
The engineers who built careers in technical writing as a primary function face the most severe restructuring pressure of any occupation in the knowledge-work technology sector. The transition pathways for this group require moving to roles that combine domain expertise with human communication skills in contexts where the audience and purpose create accountability requirements that AI cannot currently satisfy: customer advisory roles, technical sales engineering, developer relations, and product management are all potential landing zones.
Legacy Systems Maintenance - The COBOL Paradox
Legacy systems maintenance - particularly work involving COBOL, mainframe systems, older Java versions, and pre-cloud enterprise software - has historically been one of the most stable categories of technology employment because the rarity of the expertise commanded substantial premium. As the previous article in this series documented, IBM lost approximately $40 billion in market value in a single session when Anthropic announced Claude Code’s COBOL translation capabilities.
The paradox for engineers in this space is nuanced. Claude Code cannot fully replace the expertise of a seasoned mainframe engineer who understands the business logic embedded in decades of COBOL code, the risk management of live production systems running trillions of dollars in daily transactions, and the regulatory and compliance context of the industries those systems serve. The translation of code from COBOL to modern languages is one dimension of mainframe work; the safe, auditable, compliance-compliant execution of that transition in a regulated environment is a different and much harder problem.
However, the marginal economics of legacy systems maintenance are changing. The premium rate that COBOL specialists could command from clients who had no alternative is being compressed by the existence of AI translation tools that, even if imperfect, shift the negotiating dynamic. The clients who were previously captive to specialist expertise now have a credible alternative, and that alternative reduces their willingness to pay at the level the scarcity previously justified.
For engineers in legacy systems, the strategic response is not to compete with AI on code translation - that is a losing race over the medium term. It is to reposition around the dimensions of the work that AI cannot credibly perform: regulatory compliance management, risk governance for critical system transitions, and the human accountability that regulated industries require when modifying systems that process financial, health, or government data.
Part Two: The Growing Roles - Where the Genuine Demand Is
The same forces that are eliminating certain categories of IT employment are creating enormous demand in adjacent and new categories. This section documents those categories with verified salary data and the specific skills that drive the premium compensation within each.
AI/ML Engineering - The Top of the Compensation Distribution
The compensation premium for AI and machine learning engineering skills in 2026 is among the largest in the history of the technology labour market. The average AI engineer salary in the United States was $184,757 in base pay as of March 2026, with total compensation averaging $211,243 including bonus and equity. Senior AI engineers at major technology companies frequently report total compensation exceeding $300,000, and at the top end of the distribution - senior researchers and staff engineers at Anthropic, OpenAI, Google DeepMind, and the leading AI-first companies - total compensation exceeds $500,000 with large equity components.
The specific skills commanding the highest premiums within AI engineering are:
LLM fine-tuning using techniques including LoRA, QLoRA, instruction tuning, and RLHF averages $174,727 annually for specialists, with top performers exceeding $300,000. The demand for this skill reflects the enormous number of enterprises that have purchased frontier model API access and need engineers who can customise those models for specific use cases, industries, and compliance requirements.
MLOps and model deployment - maintaining, monitoring, versioning, and continuously improving AI models in production - averages $165,000 annually. The hiring difficulty for senior MLOps engineers is extreme: recruiter reports document search timelines of eleven or more weeks for senior positions, compared to four to six weeks for most other senior technical roles. MLOps does not get the media attention of LLM research, but it is the function that determines whether AI investments actually produce value in production, and the talent market reflects this reality.
RAG (Retrieval-Augmented Generation) architecture combines vector database management, embedding pipeline design, and LLM integration to allow AI systems to access organisation-specific data in production settings. This skill went from obscure to essential in approximately eighteen months. Every enterprise deploying AI for internal knowledge management, customer service, or decision support needs engineers who can build and maintain RAG pipelines. The engineers who have built RAG systems at production scale, with the performance and reliability required by enterprise settings, command premium rates that are growing faster than most other AI specialties.
Cloud AI infrastructure - designing and operating the cloud environments that serve AI workloads - averages $135,000 to $175,000 depending on seniority and the complexity of the environments managed. This role combines traditional cloud architecture skills with AI-specific requirements including GPU cluster management, autoscaling for variable AI workloads, and observability for machine learning systems.
Cybersecurity Engineering - The Counter-Wave
As AI tools expand the attack surface of every enterprise that deploys them, and as the sophistication of AI-enabled cyberattacks increases in parallel with the capabilities of defensive AI tools, cybersecurity engineering has become one of the most resilient and fastest-growing specialties in the technology sector. While the broader software engineering market has been contracting, cybersecurity hiring has remained robust or grown across the current cycle.
The demand drivers are structural and accelerating rather than cyclical:
Every enterprise AI deployment creates new attack surfaces. AI systems have specific vulnerabilities - prompt injection, model poisoning, data extraction through API exploitation, adversarial examples - that require specialised security engineering to address. The growing population of enterprises deploying AI creates a growing population of security engineers needed to protect those deployments.
The agentic AI systems being deployed through tools like Claude Code and Cowork create security challenges that are qualitatively different from those of traditional software. When an AI agent can autonomously execute code, access files, and make API calls, the blast radius of a compromised or manipulated agent is much larger than a compromised traditional software application. Designing security architectures for agentic systems is a new and specialised skill with very limited supply.
National security and defence investment in cybersecurity capabilities is growing rapidly, creating a government and defence contracting market for cybersecurity engineers that is specifically accessible to US citizens and security clearance holders. This intersection of civilian and government demand creates a salary floor that is considerably higher than in previous cycles.
The salary range for cybersecurity engineers in 2026 runs from approximately $120,000 for entry to mid-level positions to $200,000 or more for senior practitioners with specialised expertise in cloud security, AI security, or cleared government work. The IEEE’s 2026 salary outlook places cybersecurity in the highest salary growth category among all technology specialisations.
Data Engineering for AI - The Infrastructure Behind the Models
Data engineering has always been important in the technology sector, but the AI era has transformed it from a supporting function to a critical bottleneck. Every AI system requires data - for training, for retrieval-augmented generation, for continuous learning, and for performance monitoring. The engineers who design, build, and maintain the data pipelines, data warehouses, and data quality systems that feed AI applications are in extremely high demand.
Robert Half’s 2026 salary data shows that data engineers working on AI initiatives sit at the top of their internal pay bands, averaging approximately $153,750 in base salary with total compensation frequently exceeding $185,000. The IEEE salary outlook documents a 4.1% year-over-year increase for data-focused engineering roles, among the highest in the technology sector.
The specific data engineering skills most valued in the AI context include: experience with vector databases (Pinecone, Weaviate, Chroma, pgvector) for RAG pipelines; knowledge of data quality frameworks for AI training data; experience with streaming data architectures for real-time AI applications; and proficiency with the cloud data platforms (Snowflake, Databricks, BigQuery) that are the infrastructure layer for most enterprise AI deployments.
Data engineers with experience on AI-specific projects - particularly those who have built the data infrastructure for production machine learning or LLM applications - are among the most actively recruited technical workers in the current market. Their skills are sufficiently specialised that recruiting timelines are long and offer rejection rates for unsatisfying packages are high.
AI Safety and Governance - The Policy-Technical Intersection
One of the fastest-growing emerging categories in the AI employment landscape is the intersection of technical AI skills and policy, ethics, governance, and safety work. This is not primarily about academic research (though that is part of it) but about the practical enterprise need to ensure that AI systems are auditable, compliant, non-discriminatory, explainable, and safe for deployment in the contexts where they are used.
The EU AI Act, which is now in partial effect, creates specific compliance requirements for high-risk AI systems deployed in European markets. US companies deploying AI in financial services, healthcare, education, employment decisions, and critical infrastructure face an increasingly complex regulatory environment that requires people who understand both the technical capabilities and limitations of AI systems and the regulatory frameworks governing their use.
Roles in this category include AI compliance officer, AI ethics engineer, AI risk manager, responsible AI practice lead, and AI governance consultant. These are not well-defined occupational categories with established salary benchmarks in the way that software engineer or data engineer are, because the field is emerging quickly. Compensation is typically in the $130,000 to $200,000 range at established enterprises, with AI safety researchers at frontier labs earning considerably more.
The critical skill combination for this category is technical AI proficiency combined with regulatory knowledge, risk management frameworks, and communication skills that allow complex AI behaviour to be explained to non-technical stakeholders. This combination is rare because the technical AI community and the policy and risk community have historically had minimal overlap. Engineers who deliberately build both dimensions - technical competence with AI systems and policy/regulatory understanding - are positioned in a market segment with genuinely limited supply and growing demand.
Developer Relations and Technical Evangelism
As AI tools become the primary interface through which developers interact with platforms, the role of developer relations (DevRel) has become more strategic and technically demanding than ever. Companies including Anthropic, OpenAI, Google Cloud, AWS, and the rapidly growing ecosystem of AI-native startups are actively hiring senior technical professionals to serve as bridges between their AI platforms and the developer communities that build on them.
The DevRel role in the AI era is not the conference-speaking, blog-writing function it was in earlier technology cycles. It requires deep technical understanding of AI system capabilities, the ability to build compelling demonstration applications using the company’s AI tools, and the communication skills to help developers understand and adopt new AI capabilities in ways that are genuinely useful rather than promotional.
Compensation for AI-focused DevRel roles ranges from $140,000 to $200,000 at established companies, with equity upside at earlier-stage AI startups that can significantly exceed those figures. The role is particularly well-suited to engineers who enjoy teaching and communication, have built real applications using AI tools, and have a public presence through technical writing, open-source contributions, or speaking engagements.
AI-Native Startups and Early-Stage Companies
While the large tech company market for traditional software engineering roles is contracting, the venture-funded AI startup ecosystem is at one of its most active hiring phases in history. AI startup funding in 2025 and 2026 has reached record levels, with capital flowing into applications of frontier AI models across healthcare, legal, financial services, education, logistics, and defence.
These startups are not primarily hiring for the roles that are contracting at large companies. They are hiring engineers who can build AI-native products: systems designed from the ground up around LLM capabilities rather than systems that have had AI bolted on. The skills most valued at AI startups are a combination of strong software engineering fundamentals, deep familiarity with AI development frameworks, product thinking that integrates AI capabilities into user experiences, and the startup flexibility to operate effectively in fast-moving, resource-constrained environments.
Total compensation at well-funded AI startups can match or exceed Big Tech for senior engineers when equity is included. The risk profile is different - startup equity may produce nothing or it may produce life-changing wealth - but for engineers who have the skills and the appetite for the risk, the AI startup market represents genuine opportunities that are growing while the large tech market is contracting.
Part Three: The US Citizen Advantage - Pathways Unavailable to Visa Holders
This section covers the most strategically underutilised aspect of US engineer citizenship in the current market. The roles and career pathways described here are either exclusively available to US citizens, preferentially accessible through faster hiring processes, or provide substantially higher compensation premium due to their citizenship requirements.
Security Clearances and What They Unlock
A US security clearance is, in the current market, one of the most valuable career assets a technology engineer can hold. The cleared technology market - government agencies, defence contractors, intelligence community contractors, and national security technology companies - is one of the most active hiring environments for technical workers, and it is almost entirely inaccessible to non-citizens.
The three tiers of security clearance that matter most for technology engineers are:
Secret clearance is the baseline for most government and defence contractor technology work. It requires US citizenship, a background investigation covering personal history, finances, foreign contacts, and criminal record, and ongoing compliance with security requirements. Processing times currently run approximately nine to twelve months for a typical candidate, though the process can be faster for candidates with straightforward backgrounds.
Top Secret clearance unlocks a significantly larger range of government and intelligence community work. The background investigation is more intensive, covering the same areas as Secret but with greater depth, and adjudication takes longer. The salary premium for cleared engineers with TS versus Secret clearance is substantial, particularly for positions in the intelligence community.
Top Secret / Sensitive Compartmented Information (TS/SCI) is the highest standard clearance category commonly required for technology work and is required for work with intelligence agencies, special operations command, and the most sensitive national security technology programmes. The investigation and adjudication process is the most intensive, but the compensation premium over the uncleared market is the largest.
The cleared technology market pays a consistent premium over the uncleared commercial market for equivalent technical skills. For a software engineer with strong AI and cloud skills, the cleared market premium is typically 15% to 30% over equivalent commercial positions, with the premium larger for less common specialisations. For cybersecurity engineers, the cleared premium can be 25% to 40%.
Beyond the salary premium, cleared positions offer a form of employment stability that is largely absent from the commercial tech sector in the current cycle. Government and defence contracts have multi-year terms, budget certainty that commercial enterprise spending lacks, and hiring pipelines that are not subject to the quarterly earnings pressure that drives commercial tech company layoffs.
Defence Technology - The Sector Growing While Tech Contracts
The US Department of Defense has committed to massive investment in AI capabilities, with the current administration’s AI national security strategy directing billions in spending toward defence AI applications. The DoD’s AI investment strategy, the Joint AI Centre’s programmes, and the proliferating number of defence AI companies founded by veterans of Anthropic, Palantir, and other frontier AI firms represent a growing employment ecosystem that is specifically accessible to US citizens.
Defence technology companies including Palantir, Anduril, Shield AI, Scale AI (government division), C3 AI (government), and dozens of smaller companies are actively hiring engineers with AI and software skills for work that requires US citizenship or security clearance.
The intersection of AI capability and national security creates a specific technical profile in high demand: engineers who can apply frontier AI capabilities (LLMs, computer vision, multi-modal models) within the security constraints, reliability requirements, and compliance frameworks of national security applications. This is harder than it sounds. Deploying AI in an environment where model behaviour must be auditable, where failure modes have serious consequences, and where the adversarial threat includes sophisticated nation-state actors requires a different engineering discipline than deploying a consumer AI feature.
Palantir’s AI Platform for defence has been widely adopted within the US national security community. Anduril’s autonomous systems work requires AI engineers with both the technical AI skills and the domain understanding of military applications. These companies pay competitively with commercial Big Tech - total compensation packages at Palantir and Anduril for senior AI engineers are in the $200,000 to $350,000+ range - and they have access to funding streams (government contracts) that are not subject to the same commercial market pressures driving Big Tech layoffs.
For engineers who are interested in defence technology, the entry pathway typically involves either applying directly to defence AI companies that do not require pre-existing clearances (they sponsor clearances for strong candidates) or pursuing government positions that provide clearance while building experience in the sector.
Federal Government Technology Roles
The US federal government is one of the largest technology employers in the world, and federal technology positions are exclusively available to US citizens. The government’s technology modernisation agenda - moving legacy federal systems to cloud infrastructure, deploying AI tools for government services, and building cybersecurity capabilities to protect critical infrastructure - has created sustained demand for technology engineers willing to navigate the hiring process.
Federal technology salaries are lower than Big Tech in base pay but have become more competitive in recent years through special pay authorities for AI, cybersecurity, and cloud specialists. The General Schedule (GS) pay scale goes up to approximately $196,000 for senior technical positions in high-cost cities under GS-15 with locality pay adjustments. Senior IT specialists working on classified programmes through the Senior Executive Service (SES) can earn more. Special Pay Authorities for cybersecurity and AI specialists at certain agencies can push compensation to $250,000 or above.
The non-monetary benefits of federal technology employment are substantial and are particularly relevant in a market characterised by private sector layoff risk. Federal government technology workers have:
Employment stability that is dramatically stronger than commercial tech. Federal technology employees are subject to different reduction-in-force rules than private sector workers, and involuntary layoffs of career employees are rare and procedurally complex.
Defined-benefit pension programmes under FERS (Federal Employees Retirement System) that provide retirement income security absent from the 401k-only commercial sector. For engineers who have watched colleagues lose RSU value in volatile equity markets, the predictability of a defined benefit pension is meaningful.
Healthcare benefits through the Federal Employees Health Benefits programme, which provides one of the broadest sets of healthcare options available in the United States.
Student loan forgiveness through the Public Service Loan Forgiveness programme, which has particular relevance for engineers with undergraduate or graduate debt from technical programmes.
The hiring process for federal technology positions is slower and more complex than commercial hiring - the average time from job posting to offer in federal technology is three to six months versus two to four weeks in commercial tech. But for engineers who are not in an immediate financial emergency and who are thinking strategically about the next five to ten years rather than the next ninety days, the federal market is worth the investment of navigating its hiring process.
Intelligence Community Technology Work
The Intelligence Community (IC) - comprising the CIA, NSA, DIA, NGA, NRO, and associated agencies - employs substantial technology workforces that are exclusively accessible to US citizens with the appropriate clearances. IC technology work is among the most technically sophisticated and most generously compensated in the government sector.
The NSA’s Cybersecurity Collaboration Centre, the CIA’s venture arm In-Q-Tel, the NGA’s geospatial AI programmes, and the NSA’s signals intelligence processing work all require technology engineers with skills in machine learning, data engineering, software development, and cybersecurity. The compensation for IC technical workers at senior levels is competitive with the cleared commercial market and substantially above GS pay tables.
The significant constraint of IC work is the clearance requirement. Obtaining the highest clearances required for IC work takes longer than clearing for commercial defence contractors, and the lifestyle restrictions associated with clearances at these levels - restrictions on foreign travel, foreign contacts, and public communications - are more significant. For engineers whose professional lives have been internationally connected, the clearance process requires careful evaluation of what they are willing to disclose and what restrictions they can live with.
The entry pathway for engineers interested in IC work typically involves starting in a cleared contractor position with a company that provides support services to IC agencies, building a track record within the cleared ecosystem, and developing the relationships and reputation that lead to more senior cleared positions.
State and Local Government Technology - The Underutilised Pathway
Below the federal level, state and local government technology programmes represent an often-overlooked employment market for US citizen engineers. Every major US city and state is investing in digital government services, data analytics, AI tools for public service delivery, and cybersecurity infrastructure for critical systems.
The City of New York’s Technology department employs thousands of technology workers. California’s Office of Digital Innovation is one of the most active government technology offices in the country. Texas has invested heavily in state technology infrastructure. The technology salaries at senior levels in large state governments approach federal government levels, and the employment stability and benefit structures are comparable.
State and local government technology is not typically as competitive on pure salary as the commercial market for AI specialists. But for engineers who prioritise geographic stability, work-life balance, defined benefit retirement, or simply the mission-driven satisfaction of building technology that serves the public, these positions represent a genuine career pathway that does not exist for non-citizen engineers in the same form.
Part Four: The Geographic Map - Where to Be and Where to Move
The geographic distribution of AI-era technology employment is shifting from the previous pattern, and understanding where demand is concentrating versus where it is contracting matters for career decisions.
The Bay Area: Still the Highest Salaries, Changed Dynamics
The San Francisco Bay Area remains the highest-compensation technology market in the United States in 2026. The average AI engineer salary in San Francisco specifically is approximately $212,859 according to Glassdoor, materially above the national average. For engineers working at the frontier AI labs - Anthropic, OpenAI, Google DeepMind - and at Big Tech with large AI investments, the Bay Area continues to provide total compensation that is not matched elsewhere.
What has changed is the character of Bay Area technology employment. The general software engineering job market is contracting relative to its 2021-2022 peak. The office vacancy rate of 36.7% reflects the departure of headcount from the market. But the AI-specific employment market is growing, and it is concentrated in the Bay Area more than any other geography because that is where the AI labs, the AI-native startups, and the AI investment ecosystem are concentrated.
For engineers with strong AI skills who can compete for positions at frontier AI companies and AI-native startups, the Bay Area remains the highest-return geography. For engineers whose skills are in more traditional software engineering functions that are being displaced, the Bay Area is simultaneously the highest-cost and one of the most affected markets.
The practical guidance is: if your skills are in the growing categories (AI engineering, MLOps, AI security, AI infrastructure), the Bay Area premium is worth the cost of living. If your skills are primarily in contracting categories, the Bay Area cost of living combined with a harder local job market is a compounding disadvantage.
Seattle: AWS and the Cloud AI Hub
Seattle is the second-highest-compensation market for technology engineers in the United States. Amazon’s headquarters and the massive AWS presence create a specific employment ecosystem that is currently undergoing significant restructuring (Amazon’s 30,000 cuts are concentrated here) but also significant investment in AI capabilities.
The specific opportunity in Seattle is at the intersection of cloud infrastructure and AI: AWS Bedrock engineers, cloud AI architects, and data engineers working on AI infrastructure at AWS-scale. These are roles where the technical skill requirements are high, the compensation is competitive with the Bay Area, and the employment base is backed by Amazon’s continuing massive investment in AI infrastructure.
Microsoft’s Redmond campus, also in the greater Seattle area, similarly offers AI-adjacent opportunities through Azure AI services, Copilot development, and the AI infrastructure that supports Microsoft’s enterprise AI portfolio. The July 2025 Microsoft layoffs were real, but Microsoft continues to hire aggressively for AI-native engineering roles that require different skills from those eliminated.
For engineers considering Seattle, the strategic calculus is similar to the Bay Area: strong AI skills make Seattle an excellent market, while traditional software engineering skills face a harder job market in a city whose primary tech employers are cutting those roles.
Austin and Denver - The Mid-Tier Growth Markets
Austin and Denver have established themselves as the primary second-tier technology employment markets in the United States over the last five years. The migration of technology workers from San Francisco during the pandemic years, combined with significant technology employer investment in these cities, created labour markets that are now substantial enough to sustain careers across a range of technology specialisations.
For engineers who are willing to accept a salary 10% to 20% below Bay Area levels in exchange for dramatically lower cost of living, both cities offer viable career pathways in AI engineering, cloud architecture, cybersecurity, and data engineering. The cost of living adjustment often produces higher net financial outcomes for engineers outside the Bay Area even with nominally lower salaries - a $180,000 salary in Austin goes considerably further than a $220,000 salary in San Francisco.
Specific to Austin, the presence of large Dell, Amazon, and Tesla engineering operations, the growing number of AI startups, and the University of Texas’s technology research community create a diverse employment base. Specific to Denver, the concentration of aerospace and defence contractors - Lockheed Martin, Raytheon, Northrop Grumman, Boeing - creates a specific citizenship-gated opportunity in cleared technology work that is described in more detail in the defence section.
Washington DC and Northern Virginia - The Cleared Tech Hub
The Washington DC metro area, encompassing Northern Virginia and suburban Maryland, is the most important geography for the citizenship-gated career pathways described in the previous section. The concentration of federal agencies, intelligence community headquarters, and the thousands of defence and government contractors that support them creates the largest cleared technology employment market in the world.
Northern Virginia - specifically the corridor running through Tysons Corner, Reston, and Arlington - is where Booz Allen Hamilton, Leidos, SAIC, ManTech, Peraton, and dozens of other major defence contractors are headquartered. Amazon’s HQ2 in Arlington adds a major commercial tech presence. The result is a technology employment market that is distinctly different from Silicon Valley: more stable, more citizenship-gated, more security-clearance-dependent, and with a floor set by government contract rates rather than by commercial market volatility.
For engineers who are US citizens and are considering the cleared pathway, Northern Virginia offers the broadest exposure to cleared technology work in the shortest timeframe. The networking density of the cleared tech community in this geography accelerates the process of building relationships, finding positions, and eventually sponsoring clearance applications.
The salary levels in the DC area are below the Bay Area for non-cleared positions but competitive for cleared work. Total compensation for cleared senior AI or cybersecurity engineers in Northern Virginia typically runs $180,000 to $280,000 including benefits, which combined with the area’s lower housing costs (relative to San Francisco) and the employment stability of government contracting produces strong career economics.
Remote Work - The Option That Still Exists
One of the enduring legacies of the pandemic period is that remote work has become a structural feature of technology employment rather than an exception. For engineers who want to live outside major tech hubs - whether for family reasons, quality of life, cost of living, or geographic preference - the remote work option remains available in many of the high-demand role categories.
The roles most compatible with remote work in the current market are: AI/ML engineering (production work, not research), data engineering, cloud architecture, cybersecurity engineering (outside cleared programmes, which typically require on-site work), DevOps and MLOps, and developer relations. Roles that are harder to do remotely include those requiring physical access to clearance facilities, certain collaboration-intensive product roles, and work in organisations that have reversed their remote policies.
The compensation dynamics of remote work have evolved. In 2022, many remote engineers negotiated Bay Area salaries for positions in lower-cost markets. By 2026, most major employers have moved to location-adjusted remote salaries, meaning a remote engineer in Des Moines earns less than a San Francisco-based engineer in the same role. The location adjustment varies: 10% to 25% below the Bay Area is typical, though some roles have smaller adjustments.
For engineers whose primary goal is maximum total career financial outcome, remote work at a location-adjusted salary often still provides better net results than living in a high-cost tech hub on a nominally higher salary. The decision depends on the individual’s cost structure, family situation, and career stage.
Part Five: The Skills Retraining Roadmap - What to Learn, When, and How
This section provides specific, time-bounded guidance on skill acquisition for engineers at different starting points. The goal is not to produce a comprehensive curriculum but to identify the specific skills with the highest ROI in the current market, the realistic timelines for achieving marketable proficiency, and the resources with the best track record for producing that proficiency.
The Foundational Layer: What Cannot Be Shortcut
Before describing the specific skills to acquire, it is necessary to establish what cannot be shortcut. The engineers who are most successfully navigating the AI transition share a common foundation that makes their skill acquisition faster and more credible to employers:
Strong computer science fundamentals - data structures, algorithms, system design, networking, and operating systems - remain the foundation on which all other technical skills are built. The engineers most at risk from AI displacement are not those with strong fundamentals who are missing specific AI skills. They are those with weak fundamentals who have been able to paper over them with specific tool familiarity. AI tools are very effective at amplifying strong fundamentals and very ineffective at compensating for weak ones.
Software engineering practices - version control, code review, testing, CI/CD, and software architecture - remain essential and are not being displaced by AI tools but are being changed by them. Understanding how to build and maintain production software systems, evaluate code quality, and design maintainable architectures is the context within which AI tools are used effectively.
Cloud platform fundamentals across at least one of AWS, Google Cloud, or Azure are essentially table stakes for technology employment in 2026. The specific AI-era cloud skills are layered on top of this foundation, not substitutes for it.
If you have any gaps in these foundational areas, addressing them should precede or run in parallel with the AI-specific skill acquisition described below. An engineer with strong fundamentals who knows how to use AI tools is in a strong position. An engineer who knows AI tools but cannot explain how a hash table works or design a simple distributed system is not genuinely competitive in the market this article is trying to help navigate.
Tier 1 Skills: Highest ROI, Six to Twelve Months to Marketability
These are the skills that command the clearest salary premium in the current market, have the most accessible learning paths, and can be developed to job-marketable proficiency within six to twelve months by a working engineer who is investing dedicated time.
Retrieval-Augmented Generation (RAG) Architecture
Why it matters: RAG is the most commonly required AI deployment pattern in enterprise settings. Every company that wants its AI system to know about its own data, its own customers, and its own products needs someone who can build RAG pipelines that work reliably in production.
What it involves: Vector databases (Pinecone, Weaviate, pgvector in PostgreSQL), embedding models (OpenAI embeddings, open-source alternatives), chunking and retrieval strategies, reranking, and integration with LLM inference endpoints. The production dimensions include monitoring retrieval quality, managing embedding model updates, and building evaluation frameworks for RAG output quality.
Learning path: The LlamaIndex documentation and community are the best starting point for conceptual grounding. LangChain is the other major framework. Building three to five projects that use RAG with different data types (unstructured documents, structured databases, code repositories) is the fastest path to genuine competency. The total time investment to build a portfolio of RAG projects strong enough to demonstrate in interviews is approximately three to four months of regular evening and weekend work.
Compensation impact: Engineers with documented RAG production experience report 20% to 35% salary premiums over equivalent experience engineers without it.
Claude Code and Agentic Development Frameworks
Why it matters: Employers are increasingly requiring AI code tool proficiency as a baseline. Engineers who can demonstrate sophisticated use of Claude Code, GitHub Copilot, and similar tools - including the ability to design prompts that produce reliable code, validate AI-generated code for correctness and security, and architect systems that intelligently use agentic code generation - are preferred over those who cannot.
What it involves: Practical fluency with Claude Code for complex software development tasks, understanding of when to use AI generation versus manual coding, ability to design prompts that reliably produce correct implementations, and code review skills for AI-generated code. The validation and security review dimension is particularly important and particularly undervalued by engineers who assume that AI-generated code is as reliable as code they would write themselves.
Learning path: The best learning is doing. Build real projects using Claude Code across multiple languages and problem types. Document your process, including the prompts that worked, the failures, and how you validated outputs. This portfolio is more valuable to employers than any certification.
Compensation impact: Proficiency with AI coding tools is increasingly a threshold requirement rather than a premium skill - engineers without it are excluded from consideration at companies where it is standard. Having demonstrated projects on GitHub that used AI tools effectively differentiates candidates who claim the skill from those who can prove it.
MLOps and Model Operations
Why it matters: The bottleneck in enterprise AI deployment is not getting models to work in development. It is keeping them working in production - handling model drift, managing version updates, monitoring output quality, and maintaining the infrastructure that serves model inference at scale. MLOps engineers are among the hardest positions to fill in the current market.
What it involves: ML model versioning with tools like MLflow, model serving with frameworks like BentoML or Ray Serve, monitoring with Evidently AI or Arize, CI/CD pipelines for model updates, A/B testing for model changes, and infrastructure as code for ML environments. The cloud-specific dimension involves Kubernetes for orchestration, managed ML services on AWS (SageMaker) or Google Cloud (Vertex AI), and cost optimisation for inference workloads.
Learning path: The Full Stack Deep Learning course (online) is the best starting point for production ML. Deploying a complete ML pipeline on a cloud platform, from data ingestion through model serving to monitoring, and documenting the process publicly, is the portfolio piece that demonstrates genuine competency. Timeline: six to nine months to job-marketable proficiency for an engineer with existing software engineering skills.
Compensation impact: Average $165,000, with senior MLOps roles at AI-first companies reaching $200,000 to $300,000. The scarcity of candidates with production ML operations experience maintains strong upward salary pressure.
AI Security Engineering
Why it matters: Every AI deployment creates new attack surfaces. The specific vulnerabilities of AI systems - prompt injection, model poisoning, data extraction through inference, adversarial inputs - require specialised knowledge to defend against. This skill has both strong commercial demand and growing government/defence demand.
What it involves: Understanding of AI-specific vulnerabilities (the OWASP LLM Top 10 is the standard framework), ability to design security architectures that mitigate these vulnerabilities, experience with red-teaming AI systems to identify failure modes, and knowledge of data security and privacy requirements for AI training and inference. The intersection with traditional security skills (network security, application security, identity management) creates a compound skill set with strong market value.
Learning path: The OWASP LLM Top 10 documentation is the essential starting reference. The AI security research community produces accessible papers and tools, including Garak (a framework for LLM red-teaming). Building and documenting a demonstration of an AI system vulnerability and its mitigation is the portfolio piece that proves competency in this space. Timeline: four to eight months to marketable competency for engineers with existing security backgrounds.
Compensation impact: AI security specialists earn $150,000 to $240,000 depending on seniority and whether the role is in the cleared sector. The overlap with traditional cybersecurity creates compound demand.
Tier 2 Skills: Twelve to Twenty-Four Month Investment, Highest Long-Term Value
These skills require a more substantial investment and are typically best pursued by engineers who have stabilised their immediate career situation through Tier 1 skills or who are in a position to make a longer-term career pivot.
LLM Fine-Tuning and Model Customisation
This is the highest-compensation AI engineering specialty in the current market, averaging $174,727 with top performers exceeding $300,000. The skills involved are more research-adjacent than most application engineering: understanding of transformer architectures, experience with parameter-efficient fine-tuning methods (LoRA, QLoRA, adapter layers), knowledge of instruction tuning and RLHF, and ability to evaluate model quality rigorously.
The learning path is more demanding than Tier 1 skills. A strong mathematical foundation (linear algebra, calculus, probability and statistics) is a prerequisite. The fast.ai courses provide an accessible entry point that does not require a research background. Completing the Hugging Face NLP course and fine-tuning several models on domain-specific datasets produces the portfolio evidence employers require. Timeline: twelve to eighteen months for an engineer with strong fundamentals and no prior ML background.
The compensation premium for this skill makes the investment ROI among the highest available. An engineer who transitions from a $120,000 manual QA role to an LLM fine-tuning role over an eighteen-month retraining period is making a compensation decision worth $50,000 to $150,000 annually for the remainder of their career.
Cloud Architecture with AI Specialisation
The average US cloud engineer salary in 2026 is $135,000 to $152,000 in base pay with total compensation frequently exceeding $175,000. For cloud architects with specific AI infrastructure specialisation, the range extends to $200,000 and above. The cloud architect role has the advantage of building on skills that many engineers already have (software development, networking fundamentals) while adding the architecture and design competency that commands premium rates.
The specific cloud architecture skills with the highest AI-era demand involve designing multi-cloud architectures that optimise for AI workloads: understanding how to allocate training workloads (typically on GPU-rich environments) versus inference workloads (where cost and latency optimisation are primary), designing data architectures that feed AI pipelines efficiently, and implementing observability and reliability for AI-serving infrastructure.
The certification path provides a structured learning framework: AWS Solutions Architect Professional, Google Professional Cloud Architect, and Azure Solutions Architect Expert are the industry-standard credentials. For AI-specific depth, the AWS Machine Learning Specialty, Google Professional Machine Learning Engineer, and Azure AI Engineer Associate certifications add significant market value and correlate with 10% to 15% salary premiums above the base cloud architect rate.
Quantitative AI for Finance
For engineers with existing quantitative or financial domain knowledge, the intersection of AI and quantitative finance creates one of the highest-compensation niches in the market. Hedge funds, quantitative trading firms, and investment banks building AI-assisted trading and risk management systems pay total compensation of $400,000 and above for senior engineers who combine strong ML skills with quantitative finance domain knowledge.
This pathway is only available to engineers who either already have a quantitative background or are willing to invest in building it alongside the AI skills. The learning path involves quantitative finance fundamentals (stochastic calculus, options pricing, risk management), Python-based quant finance libraries (QuantLib, pandas for financial time series, factor modelling), and the specific ML applications in finance (time series forecasting, portfolio optimisation, risk modelling, alternative data analysis).
The barrier to entry is high, but so is the compensation. For engineers with strong mathematics backgrounds who are currently in lower-paying generic software engineering roles, this pathway offers a large and lasting compensation improvement.
Certifications Worth Pursuing in 2026
Beyond the informal skill-building described above, certain certifications carry consistent market value and are worth the investment of time and money:
AWS certifications, particularly the Solutions Architect Professional and Machine Learning Specialty, are the most broadly recognised cloud credentials and correlate with 10% to 15% salary premiums. The AWS Professional path requires passing the Associate-level exam first.
Google Cloud certifications, particularly the Professional Machine Learning Engineer, are the most valued for engineers targeting the AI infrastructure space given Google’s central role in AI compute.
Azure certifications including AZ-104 (Administrator) and AI-102 (AI Engineer) are most valuable for engineers whose target employers use Microsoft’s stack - the enterprise market where Azure dominates.
The Certified Information Systems Security Professional (CISSP) remains the gold standard for cybersecurity careers and is worth pursuing for engineers moving into security. The study investment is significant (typically six to nine months of serious preparation) but the credential opens clearance-adjacent and senior security roles that are otherwise difficult to access without the designation.
The Certified Cloud Security Professional (CCSP) is the best intersection credential for engineers working at the overlap of cloud and security - a fast-growing category in the cleared and commercial markets.
Learning Resources Worth the Investment
The market for technical education has expanded dramatically alongside the demand for AI skills, and not all resources are equivalent in their return on time investment. The following are consistently rated as highest-value by practicing engineers in the current market:
Fast.ai’s Practical Deep Learning course is the best entry point for engineers with programming skills who want to build genuine ML competency without a research background. The bottom-up pedagogical approach builds intuition before formalism and produces practitioners faster than top-down academic approaches.
The Anthropic developer documentation and API guides are the most current and practical resource for working with Claude specifically. For engineers building Claude-based applications, the documentation quality is high and the examples are production-relevant.
Hugging Face courses and documentation are the standard resource for transformer-based NLP and fine-tuning workflows. The Hugging Face community (particularly the Forums and Spaces) is where practitioners share working implementations and discuss production challenges.
The Full Stack Deep Learning course (FSDL) is the best resource for the MLOps and production ML dimension. The course covers the full lifecycle from training to monitoring and is updated regularly to reflect current production practices.
Towards Data Science on Medium, while variable in quality, contains a significant volume of practitioner-authored tutorials on specific AI engineering tasks that are difficult to find in more formal resources.
The AWS, Google Cloud, and Azure documentation and hands-on labs are the authoritative resources for cloud-specific AI services. The cloud providers have invested heavily in tutorial content because they have commercial incentives to accelerate developer adoption of their AI services.
Part Six: The 90-Day Action Plan by Career Stage
The preceding sections provide the strategic framework. This section provides the specific, time-bound actions that translate that framework into career movement. Three distinct plans are provided: one for early-career engineers (zero to five years of experience), one for mid-career engineers (five to fifteen years), and one for senior engineers (fifteen or more years).
The Early-Career Engineer (Zero to Five Years): Weeks One Through Twelve
Week One: Assessment and Positioning
Begin with an honest skills inventory that maps your current competencies against the three categories described earlier: functions being eliminated, functions growing, and citizenship-gated pathways. Be specific: “Python programming” is not a marketable skill in isolation; “Python data pipeline development with Airflow and dbt, deployed on AWS, with documented production experience” is a marketable skill.
If your current role falls primarily in the eliminated categories (manual QA, routine code maintenance, basic data entry engineering), accept the urgency of the situation and plan the transition rather than waiting for market pressure to force it.
Document your existing projects in detail. Create or update your GitHub profile to ensure that every project you have built is publicly visible with clear README documentation that explains what the project does, why you built it, and what decisions you made. For early-career engineers, the GitHub profile is often the primary evidence base for interview screening.
Apply for roles in your target category immediately, even if you do not yet have all the skills. The application process provides feedback about what employers are actually looking for that is more useful than any analysis article, and early interviews produce information that refines your skill investment direction.
Weeks Two Through Four: First Skill Sprint
Choose one Tier 1 skill from the list above and begin concentrated work on it. The selection criterion should be the overlap between your existing skills and the skill being acquired: an engineer with existing database experience should prioritise RAG over MLOps; an engineer with existing security experience should prioritise AI security; an engineer who already uses AI coding tools heavily should prioritise demonstrating that proficiency in a portfolio project rather than starting something entirely new.
The week two to four skill sprint should produce a working demonstration project by the end of week four. The project does not need to be production-grade; it needs to be visible on GitHub with documentation that explains what you built and the decisions you made. The act of building and documenting forces the depth of learning that reading and watching tutorials does not.
Spend fifteen minutes daily on LinkedIn, engaging with content in the AI engineering community, commenting substantively on posts from engineers working in your target roles, and updating your profile to reflect your current learning direction.
Weeks Five Through Eight: Network and Application Investment
Networking is the highest-ROI activity for early-career engineers who feel like they lack the experience to network effectively. The reality is that the AI engineering community is actively welcoming to engaged early-career practitioners who demonstrate genuine technical interest and are building real things.
Attend every local meetup, virtual conference, or workshop related to AI engineering, MLOps, or your target specialisation. The specific events worth prioritising are those organised by the cloud providers (AWS, Google Cloud, and Azure all run free practitioner-level events), the Hugging Face community events, and local ML practitioner groups (check Meetup.com for your city).
Cold outreach on LinkedIn to engineers working in your target role at companies you want to work for has a materially higher response rate than most people expect, particularly if the message is specific, brief, and asks for something small (thirty minutes to learn about their role, not an introduction to the hiring manager).
Continue the portfolio project from weeks two through four and begin a second project that demonstrates a complementary skill or a more complex version of the first.
Weeks Nine Through Twelve: Amplify and Apply
The first resume-ready portfolio piece should be complete and demonstrable by week nine. Use it actively in your application process. For each application, customise the cover letter to connect your specific portfolio projects to the specific requirements of the role.
If you are still in your current job during this period, investigate internal opportunities. Many large technology companies have internal mobility programmes that allow employees to move to AI-adjacent teams before external job searching. This path is often faster and involves less disruption to compensation and benefits than an external move.
If you are currently unemployed, prioritise the application process alongside the skill building. The skill building is important, but income is more immediately important than resume-optimisation. Contract and freelance work through platforms including Toptal, Arc, and Upwork that specifically serve technical professionals can provide income while building portfolio evidence in your target area.
Request informational conversations at every opportunity. The engineers who are most effective at career transitions in the current market are those who have five to ten informational conversations per week with people working in their target roles. These conversations provide specific, current information about what employers actually want that is more useful than any general guide.
The Mid-Career Engineer (Five to Fifteen Years): A Different Strategy
Mid-career engineers face a different set of constraints and opportunities from early-career engineers. They typically have more financial cushion, more professional credibility, and more specific domain expertise. They also have more to lose in terms of compensation level, title, and professional identity, and they face the specific challenge that their most valuable existing skills may be in the categories under pressure.
Weeks One Through Two: Honest Career Mapping
The mid-career engineer’s first action should be an honest assessment of what they have that is genuinely defensible versus what is at risk. The framework is: domain expertise (knowledge of a specific industry or problem domain) is typically defensible and increasingly valuable as AI tools make technical implementation less differentiating. Technical implementation skills in high-demand AI categories are defensible. Technical implementation skills in categories being automated are not defensible as standalone career capital, though they may still have value when combined with domain expertise.
A mid-career engineer who has spent ten years working in healthcare IT has domain expertise that is both genuinely valuable and hard to acquire. If their technical skills are primarily in manual QA or legacy Java maintenance, the combination of healthcare domain expertise plus newly acquired AI integration skills creates a genuinely valuable profile: someone who can deploy AI tools in healthcare contexts with the domain knowledge to do it correctly in a regulated environment. The domain expertise makes the AI skills more valuable; the AI skills make the domain expertise more deployable.
This framework - domain expertise plus AI skills - is the mid-career engineer’s most defensible positioning in the current market, and it requires less pure technical retraining than the early-career path because the domain component is already built.
Weeks Three Through Six: Identify the Premium Combination
Using the domain expertise plus AI skills framework, identify the specific combination that represents your highest-value positioning. Examples:
A mid-career engineer with strong financial services domain expertise and existing Python skills who adds RAG architecture and LLM API integration skills becomes a “fintech AI engineer” - someone who can build AI tools for financial applications with the regulatory and domain knowledge to do it correctly. This profile is in very high demand at banks, fintech companies, and financial technology startups.
A mid-career engineer with healthcare IT background who adds AI integration skills and HIPAA compliance knowledge for AI systems becomes a “healthcare AI engineer” - a profile sought by hospital systems, healthcare technology companies, and the growing category of AI health companies.
A mid-career engineer with manufacturing or industrial engineering background who adds computer vision and IoT integration skills becomes positioned for the growing industrial AI market, which is one of the fastest-growing segments of AI application development.
Identify your domain, map the AI skills that complement it most directly, and focus your retraining investment on that specific combination rather than pursuing generic AI engineering skills.
Weeks Seven Through Ten: Management or Technical IC Decision
Mid-career engineers face a career path decision that early-career engineers do not: the choice between a management track (engineering manager, director of engineering, VP of engineering) and a technical individual contributor track (staff engineer, principal engineer, distinguished engineer). The AI era affects this decision in specific ways.
The management track in traditional software engineering has been compressed by the AI-driven elimination of middle management layers at companies like Amazon and Microsoft. The companies cutting “anti-bureaucracy” layers are specifically targeting the roles that mid-career engineers transition into when they move into management. For mid-career engineers considering the management track, the key question is whether they are moving toward strategic management (which adds judgment, vision, and organisational leadership that AI cannot replace) or toward coordination management (which is being automated).
The technical IC track is gaining value relative to management in the AI era, particularly at the staff and principal engineer level. These roles - which involve architectural decisions, technical strategy, and the kind of systems-level thinking that AI tools cannot currently perform - are increasingly important as companies rely on smaller technical teams doing more with AI assistance. A principal engineer at an AI-first company who can set the technical direction for AI-enabled systems development commands $250,000 to $400,000 and is in genuine demand.
If you are considering the management track, ensure your management aspirations are toward strategic roles rather than coordination roles. If you are considering the technical IC track, identify the specific technical domain where you want to build staff-level depth.
Weeks Eleven Through Twelve: The Security Clearance Decision
For mid-career engineers who are US citizens and who have not previously explored the cleared sector, weeks eleven and twelve are the right time to seriously evaluate this pathway. The cleared sector decision involves several considerations:
Background: Security clearances require disclosure of financial history, foreign contacts, foreign travel, and personal history that some engineers may not be comfortable disclosing or that may be complicated by their personal circumstances. Evaluating whether your background makes clearance achievable is a necessary first step.
Lifestyle: Cleared work typically requires being physically present at a cleared facility for at least part of the work, which may conflict with fully remote work arrangements. The highest-clearance positions have the most significant restrictions on foreign travel, foreign contacts, and public communications.
Timeline: Initiating a clearance process through a cleared employer who will sponsor you takes nine to eighteen months to complete. This is not a ninety-day action but a twelve to twenty-four month career move.
If the background and lifestyle considerations are manageable, the financial and stability case for the cleared pathway for mid-career engineers is strong. The combination of base salary premium, benefit package quality, employment stability, and long-term career trajectory in the growing defence AI sector represents one of the better medium-term career investments available to US citizen engineers in the current market.
The Senior Engineer (Fifteen or More Years): Repositioning from Value to Strategy
Senior engineers with fifteen or more years of experience face a paradox in the current market. Their technical skills may include significant portions that are in the automated category, but their accumulated experience, judgment, and credibility represent exactly the kind of human capital that is most difficult to replace with AI. The strategic challenge is repositioning from being valued primarily for technical implementation to being valued primarily for technical judgment, strategic direction, and the human accountability that enterprise AI deployments require.
Identifying Your Authentic Senior Value
The senior engineer’s most defensible career capital falls into several categories:
Systems-level thinking about architecture, scale, and technical risk that emerges from having designed and built systems across multiple technology generations. An engineer who was there when microservices replaced monoliths, who understands why distributed systems fail in specific ways, and who can evaluate the second and third-order consequences of architectural decisions has knowledge that is not easily acquired and not easily replicated.
Organisational and cross-functional credibility built over years of working with product managers, business stakeholders, and non-technical leadership to translate technical constraints into business decisions and business requirements into technical specifications.
Domain expertise in specific industries, regulatory environments, or problem categories that combines technical depth with business context. The engineer who has spent fifteen years in healthcare IT does not just know the technology; they know the workflows, the regulatory environment, the organisational dynamics of healthcare organisations, and the failure modes that matter in that specific context.
Technical leadership and mentoring capability - the ability to set technical direction for teams, develop other engineers, and maintain quality standards across a complex organisation.
These forms of capital are not rendered worthless by AI. They are made more valuable in a context where junior engineering work is increasingly AI-assisted and where the primary constraint on AI-enabled teams is the judgment, direction, and accountability provided by senior technical leaders.
The Consulting and Advisory Pathway
For senior engineers with fifteen or more years of experience, independent consulting or advisory work is a pathway that capitalises on accumulated expertise without the vulnerability of depending on a single employer who may restructure.
The market for senior engineering advisors has grown alongside the AI transition, as companies that are restructuring their engineering organisations simultaneously need external guidance on how to execute those restructurings effectively. A former Amazon or Microsoft engineering director who understands AI-enabled engineering organisations at scale has market value as an advisor to companies attempting to build those capabilities.
Setting up an independent practice requires several investments: building a public professional presence through writing, speaking, or advisory roles that establishes credibility with potential clients; creating a clear service offering that articulates the specific problems you solve and for whom; and building a pipeline of client relationships through your existing professional network.
The economics of senior engineering consulting can be excellent. Day rates of $2,000 to $5,000 are achievable for advisors with genuine expertise and established credibility. Project-based engagements can produce $50,000 to $200,000 for work that takes weeks rather than months. The variability is higher than employment, but the ceiling is also substantially higher.
Entrepreneurship in the AI Era
For senior engineers with strong technical skills, deep domain expertise, and entrepreneurial inclinations, the current AI era represents one of the best periods in the history of technology to start a company. The availability of powerful AI tools that can dramatically accelerate product development, combined with the venture capital appetite for AI-native companies, creates conditions where small teams of experienced engineers can build products that would have required much larger teams in previous technology cycles.
The specific opportunities most suited to experienced senior engineers involve applying AI tools to solve problems in industries or domains where the founder’s domain expertise creates a genuine competitive advantage over teams without that expertise. Healthcare AI companies founded by engineers with clinical domain knowledge, industrial automation companies founded by engineers with manufacturing experience, legal technology companies founded by engineers with legal domain knowledge - these are the patterns where the “just apply AI to it” approach is wrong, and where genuine domain expertise creates durable competitive advantage.
Part Seven: The Financial Dimension - Managing the Transition
Career transitions take time, and time costs money. This section addresses the financial management of the transition period with the specificity that generic career advice typically omits.
Understanding Your Severance Package
For engineers who have already been laid off or who believe a layoff is imminent, understanding the severance package in detail is the most immediately financially important action. Severance packages in the technology sector typically include:
A base severance amount calculated as weeks of pay per year of service, typically capped at some maximum. The range at major tech companies runs from four weeks at the low end to twenty or more weeks for long-tenure employees. Amazon’s January 2026 packages reportedly included up to five months for long-tenure employees; Microsoft’s July 2025 packages included at least 90 days.
Benefits continuation including health insurance through COBRA or equivalent coverage for the severance period. Understanding when your health coverage ends and what the COBRA premium will be is critical because unexpected large medical expenses are among the most common financial catastrophes for unemployed workers.
RSU vesting acceleration or forfeiture. Most technology companies do not accelerate RSU vesting upon layoff, meaning that unvested stock grants are forfeited when you leave. For engineers with large unvested RSU positions, negotiating extended vesting or an accelerated vest as part of a separation agreement is worth attempting, particularly in situations where the layoff is part of a company-initiated restructuring rather than a performance action.
Outplacement services, career coaching, and training resources that are sometimes included in separation packages. These vary widely in quality but should be utilised even if only modestly useful.
The Unemployment Benefits Calculation
US unemployment insurance (UI) provides a financial floor during job transitions that is available to all US workers regardless of citizenship status. The benefits calculation varies by state but typically replaces approximately 40% to 50% of prior weekly earnings up to a state-specific maximum.
The maximum weekly UI benefit in California (as of 2026) is approximately $1,900. In New York, it is approximately $1,600. In Texas, it is approximately $1,000. For tech workers earning $150,000 or more, UI replaces only a fraction of prior income, but it provides meaningful financial support during a transition period.
Filing for UI immediately upon separation is important because the processing and first payment typically take two to three weeks after filing. Many engineers delay filing because of embarrassment or because they expect to find a new position quickly, and the delay costs them benefits they are entitled to. File immediately.
Financial Runway Planning
The median time to re-employment for laid-off tech workers has increased from 3.2 months in 2024 to 4.7 months in early 2026, and for senior engineers seeking specific AI-adjacent roles, timelines of six to nine months are realistic. Planning financial runway accordingly is the most important financial action during a transition.
The runway calculation is straightforward but requires honesty about both income (severance, UI, potential freelance) and expenses. A three-month severance package, plus UI at the maximum, plus whatever savings provide buffer, needs to cover living expenses through the expected job search duration plus a contingency. Engineers who plan for a three-month search and find themselves at month six without an offer face compounding stress that impairs the judgment and energy the job search requires.
Specific cost reduction actions worth taking during a technology job search: renegotiating or pausing subscription services; reviewing whether to exercise or hold stock options before they expire; understanding whether 401k contributions can be reduced during the transition period to improve cash flow; and evaluating whether a geographic move to a lower cost-of-living city is worth making during rather than after the search.
For engineers with significant stock-based compensation from prior employers, consulting a financial advisor about the tax implications of exercising options or selling restricted stock during a gap year is worth the advisory fee.
Part Eight: The Mental Health Dimension
No career article would be complete without addressing what the data consistently shows: job loss and prolonged job searching produce mental health impacts that are functionally impairments to the job search process itself. This is not a soft addition to a hard-nosed career guide. It is recognition that managing the psychological dimension of this transition is a career strategy requirement, not a personal wellness bonus.
What the Evidence Shows About Layoff Psychology
Research on involuntary unemployment consistently shows elevated rates of depression, anxiety, and sleep disruption that begin within weeks of job loss and persist through the transition period. For technology workers who have built significant professional identity around their roles - and whose social networks are heavily concentrated within their employer - the loss of the job involves more than financial disruption. It involves loss of daily structure, collegial relationships, professional identity, and the sense of purpose that comes from doing meaningful work.
The specific patterns that are particularly common among laid-off technology workers in the current cycle include:
Over-application followed by disengagement: Engineers who apply to a large number of positions quickly, receive rejections or silence, and then disengage from the search entirely. The initial application surge is a natural response to anxiety; the disengagement is depression masquerading as strategy.
Isolation and information echo chambers: The tendency to spend the job search period primarily online, reading layoff forums and news articles that reinforce anxiety without providing actionable information. This creates a feedback loop that amplifies fear without directing energy productively.
Perfectionism paralysis: The tendency to over-prepare for each application and each interview to the point where preparation crowds out volume. The search that produces one perfect application per week produces worse outcomes than the search that produces ten imperfect but genuine applications per week.
Identity rigidity: The resistance to considering role categories that were not part of the previous career path, motivated by a desire to preserve career status rather than find meaningful work. An engineer who refuses to consider contractor roles because they want a “real job,” or who refuses to consider companies outside their previous company’s prestige tier, is allowing identity concerns to narrow the search in ways that extend its duration unnecessarily.
Specific Actions That Help
Maintain daily structure even without an employer providing it. Wake at a consistent time, designate specific hours for job search activity, exercise daily, and end the “work day” at a defined time. The absence of externally imposed structure is a risk factor for the disengagement patterns described above, and self-imposed structure is the mitigation.
Maintain social connection actively. The professional relationships that feel least urgent to maintain during a job search are often the most valuable: the former colleague whose opinion you trust, the manager who encouraged your development, the peer who knows your work quality. These relationships require deliberate maintenance during periods when the shared context of daily work has ended.
Limit passive news consumption about layoffs. Knowing the aggregate statistics of the layoff wave does not make individual navigation of it more effective. Spending two hours reading about Amazon’s latest cuts when you left Amazon six weeks ago is not productive research. It is anxiety maintenance. The time is better spent on applications, portfolio work, or conversations with people who are actually hiring.
Seek professional support early rather than late. The stigma around mental health support is lower in the technology community than in many other professional environments, and the practical value of a few sessions with a therapist who works with career transitions is high enough that it is worth the investment regardless of whether you feel like you “really need it.”
Frequently Asked Questions
Q1: I am a US citizen on a traditional software engineering salary of $130,000. What is my realistic two-year compensation outcome if I reskill effectively?
If you are currently in a role with significant exposure to the automated categories (manual QA, routine code maintenance) and you pursue a focused twelve to eighteen month retraining into MLOps or RAG architecture, the realistic two-year compensation outcome based on current market data is $165,000 to $185,000 in base salary with total compensation of $190,000 to $220,000. That represents a 27% to 40% compensation improvement over the current baseline, which is a strong ROI on a twelve to eighteen month retraining investment. Achieving this requires documented production experience with the target skills, not just certification, and a geographic strategy that targets markets where those skills are in highest demand.
Q2: I have fifteen years of software engineering experience but no AI skills. Is it too late to make the transition?
No. The engineers most in demand in the current market are those who can combine AI skills with deep domain expertise or systems architecture understanding that only comes from years of experience. The pure AI researcher roles require research backgrounds that most experienced software engineers do not have. But the applied AI roles - the MLOps engineer who has also built and run production software systems, the AI security engineer who understands both ML vulnerabilities and enterprise security architecture, the staff engineer who can design systems that intelligently incorporate AI components - are roles where fifteen years of experience is a significant advantage, not a liability.
Q3: Should I pursue the security clearance pathway? How do I start?
If you are a US citizen with a reasonably clean background (no foreign national spouses or close family, no significant debt problems, no criminal history) and you are willing to work in Northern Virginia or the other major cleared tech hubs, the cleared pathway is worth serious consideration. The starting point is applying to cleared technology companies that sponsor clearances - Booz Allen Hamilton, Leidos, SAIC, Peraton, and many smaller defence contractors regularly hire engineers and sponsor their clearances. The sponsoring employer submits the clearance application on your behalf, and you begin work in an unclassified capacity while the investigation proceeds. Be aware that the process takes nine to eighteen months and requires honest disclosure of everything the investigation asks.
Q4: What is the salary impact of remote work versus in-person for AI engineering roles?
The location adjustment for remote AI engineering roles is typically 10% to 25% below the Bay Area equivalent, depending on the employer’s location policy. Many companies have formalized location tiers that determine remote salary levels. For engineers in lower-cost markets, the net financial outcome after location adjustment and cost of living comparison is often favourable relative to living in San Francisco on a nominally higher salary. The key question is whether the role you want is available remotely - some roles, particularly those involving cleared work, on-site lab access, or high-collaboration product teams, require physical presence.
Q5: I just got laid off. What are the first three things I should do?
File for unemployment insurance immediately - do not wait. Update your LinkedIn profile to “open to work” (the hidden version that shows only to recruiters, not your entire network, if you prefer privacy) and ensure your profile accurately reflects your most current and most AI-relevant skills. Send five to ten personalised messages to people in your professional network letting them know you are exploring new opportunities and asking for a brief conversation - not asking for a referral, just for a conversation. These three actions take less than a day and initiate the financial support, the recruiter pipeline, and the referral network that are the primary sources of job opportunities in the current market.
Q6: Is the government technology path worth the lower salary?
For engineers at certain career stages and with certain priorities, yes. The calculation depends primarily on the compensation comparison after accounting for the full benefit package, not just base salary. The Federal Employees Retirement System defined benefit pension, combined with subsidised healthcare, Federal Employees Group Life Insurance, and the FEHB programme’s breadth, adds substantial non-salary compensation that makes the total package more competitive than the base salary comparison suggests. For engineers who prioritise employment stability above marginal compensation maximisation, or who have significant student loan debt that qualifies for PSLF, or who genuinely want to work on public interest technology, the government path makes strong career sense.
Q7: Should I consider leaving the US for technology work elsewhere?
For US citizen engineers, leaving the US for technology work is typically a compensation downgrade unless the destination is a specific high-paying market (Zurich, London, Singapore) and the role is specifically in a high-paying specialisation. The US technology compensation market, even in the current disrupted state, is the highest-paying in the world for technology workers outside of a handful of specific role-geography combinations. The practical advantage of US citizenship is primarily within the US market, where citizenship unlocks the cleared and government sectors that provide the most stable and best-compensated alternatives to Big Tech in the current cycle.
Q8: Which AI specialisation has the best combination of high compensation, reasonable retraining time, and sustainable long-term demand?
MLOps is the most defensible answer. The compensation is strong ($165,000 average, with senior roles reaching $250,000 to $300,000). The retraining time for an engineer with existing software engineering skills is six to nine months to job-marketable proficiency. And the long-term demand is sustainable because the need to maintain, monitor, and continuously improve AI models in production is a function that grows with AI adoption rather than being displaced by it. MLOps is not the highest-ceiling specialty (fine-tuning research commands larger compensation peaks) but it has the most consistently strong economics across the career range from mid-level to senior.
Q9: I have both strong coding skills and strong communication skills. What is the best career path?
Developer relations and technical product management. Developer relations at AI-first companies (Anthropic, OpenAI, AWS, Google Cloud, AI-native startups) combines technical depth with public communication and community building in ways that are genuinely hard to find and strongly compensated at $140,000 to $200,000 plus equity. Technical product management, particularly for AI products, requires the ability to understand what AI systems can and cannot do technically while communicating with non-technical stakeholders about product requirements and trade-offs - a combination that is in extreme shortage in the current market.
Q10: How do I compete with laid-off senior engineers from Amazon, Google, and Microsoft who are flooding the same AI engineering job market?
By being specific where they are generic. The engineers who have been most impacted by Big Tech layoffs are, on average, engineers from large, well-resourced teams who built specific components of large systems at scale. They may have deep expertise in those specific components but limited breadth across the AI stack. Engineers who can demonstrate end-to-end AI system experience - from data pipeline through model training to production serving and monitoring - are differentiating themselves from engineers who know only the application layer or only the infrastructure layer. Building demonstrable end-to-end AI project portfolios, with all the layers visible and documented, is the competitive strategy that works against the volume of more narrowly experienced engineers in the current market.
Q11: Is the tech startup ecosystem a safer or riskier bet than large companies right now?
The answer depends on stage. Pre-product startups (seed and early series A) are riskier than large companies in terms of the probability of total company failure, but offer the most equity upside if they succeed. Series B and C AI-native startups that have found product-market fit and have two or more years of runway are in some ways more stable than large tech companies, because they are in growth mode rather than the restructuring mode that is driving Big Tech cuts. The key is evaluating the specific company’s position: runway, revenue trajectory, customer concentration, and the quality of the investors on the cap table are more important predictors of short-term employment stability than the stage label.
Q12: Is it worth relocating to Northern Virginia specifically for the cleared tech market?
If you are a US citizen who has determined that the cleared pathway makes sense for your background and preferences, and if you do not have strong roots keeping you in your current location, Northern Virginia is the most efficient geography for building a cleared tech career. The density of cleared employers, the networking community within the cleared tech ecosystem, and the breadth of the opportunity range mean that your trajectory in the cleared market will be materially faster than if you pursue cleared work from a geography with fewer cleared employers. The cost of living in Northern Virginia, while higher than interior markets, is substantially below San Francisco and Seattle, and the combination of cleared salary premiums and lower housing costs produces strong overall financial outcomes.
Q13: I’ve been in the same company for twelve years. My skills are all in their internal proprietary systems. How exposed am I?
This is one of the most genuinely vulnerable positions in the current market: deep expertise in systems that are not transferable combined with long tenure that may have been accompanied by salary growth not reflected in market rates. The risk is that your skills are not legible to other employers, your salary may be above market for the skills that are legible, and AI tools may be compressing the need for the institutional knowledge you have built. The most urgent action is to begin building market-legible skills immediately - ideally while still employed - rather than waiting for a layoff announcement to motivate the transition. Internal movement to AI-adjacent teams within your current employer, if available, is the fastest path to both building new skills and maintaining employment while you do it.
Q14: What is the realistic salary premium for an engineer who gets a security clearance?
For engineers with strong AI or cybersecurity skills who obtain a Secret clearance, the premium over equivalent uncleared positions is typically 15% to 25%. For TS clearance, the premium is 25% to 40%. For TS/SCI with polygraph (required for some intelligence community work), the premium can reach 50% above uncleared equivalents in less specialised markets like Denver or Northern Virginia. The premium reflects both the genuine scarcity of cleared engineers and the salary floor set by government contract rates, which are higher for cleared positions than the commercial rate for many equivalent roles outside high-cost tech hubs.
Q15: I am in my late forties. Is it worth the investment to retrain for AI skills, or should I be thinking about something else entirely?
The retraining investment absolutely makes sense at any age where the career years remaining exceed the retraining period by a sufficient margin. For an engineer in their mid-to-late forties with fifteen or more years of remaining career potential, an eighteen-month retraining investment that produces $40,000 to $80,000 in annual compensation improvement pays back at a multiple of ten or more over the remaining career horizon. Age discrimination is a real phenomenon in technology hiring, but it operates primarily at the screening stage and affects engineers who are not staying current. Engineers in their late forties who are demonstrably proficient with AI tools, who have recent portfolio projects, and who have strong domain expertise are genuinely competitive in the current market. The advice to “consider something else entirely” is appropriate for engineers who genuinely do not want to continue in technology, but it is not an age-driven requirement.
Conclusion: The Citizenship Advantage Is Real but Time-Sensitive
The US citizen engineer’s position in the 2026 technology labour market is more complex and more full of genuine opportunity than either the panic-inducing headlines or the reassuring platitudes suggest. The headlines are right that significant disruption is underway. The platitudes are right that the disruption creates as well as destroys. What neither provides is the specific, actionable guidance that allows individual engineers to make the decisions that actually matter.
This article has attempted to provide that guidance based on verified data rather than extrapolation from headlines. The key points bear restating in summary:
The roles being eliminated are specific: manual QA, routine code maintenance, coordination programme management, and legacy systems work without the judgment and compliance expertise that makes it defensible. Engineers in these roles have an eighteen to twenty-four month window of urgency - not an immediate crisis, but not something to defer.
The roles growing are also specific: AI/ML engineering with production credentials, MLOps, RAG architecture, AI security, data engineering for AI, and developer relations for AI platforms. These roles command compensation premiums of 20% to 60% above equivalent experience in contracting categories.
The citizenship advantage is real, specific, and currently underutilised. Clearance-eligible work, defence technology, federal technology employment, and intelligence community contracting represent a large and growing employment sector that is almost entirely inaccessible to non-citizen engineers. For engineers whose current commercial tech trajectory is under pressure, these pathways provide an alternative that is both financially attractive and employment-stable.
The geographic opportunities are distributed more widely than the traditional tech hub concentration suggests. Northern Virginia for cleared work, Austin and Denver for mid-tier market economics, and remote-compatible AI engineering roles for engineers who can demonstrate the required skills - the geography of opportunity in 2026 is broader than the geography of opportunity in 2022.
The skills retraining investment has clear, calculable ROI. An engineer who invests six to twelve months in MLOps, RAG, or AI security skills and documents that investment through portfolio projects is not making a leap of faith. They are making a decision with well-documented expected outcomes that is financially similar to a professional degree programme but faster and cheaper.
The psychological dimension is part of the career strategy, not separate from it. Managing the stress and uncertainty of the current transition with the same intentionality brought to skill investment is a requirement for effective execution, not a soft addition to a hard strategy.
The engineers who navigate this transition best will not be those who panic or those who ignore it. They will be those who read the evidence accurately, identify their specific position within it, make deliberate decisions about which opportunities to pursue, and invest the consistent effort that career transitions require without being paralysed by the uncertainty that makes consistent effort difficult.
The tools to do this work exist. The opportunities are real. The citizenship advantage is significant. The decision belongs to each engineer reading this.
This article was produced using verified data from IEEE Spectrum, Robert Half’s 2026 Salary Trends, Built In salary data, Glassdoor 2026 industry reports, Anthropic’s Economic Index and labour market research, Axiom Recruit AI engineering compensation analysis, SSi People engineering job market forecasts, and reporting from Newsweek, Rest of World, SF Standard, Fortune, and IEEE. All salary figures are for US-based positions unless noted. The career guidance represents the best available evidence-based assessment as of March 2026 in a rapidly evolving market. Individual outcomes depend on specific skills, background, geography, and execution of the strategies described. For technology career preparation resources, visit ReportMedic.
Part Nine: The Salary Negotiation Playbook for AI-Era Roles
One of the most consequential practical skills in the current market is negotiating compensation for AI-engineering roles, where the salary bands are wide, the market data is sparse, and the premium for negotiating well versus accepting the first offer is large. This section provides specific guidance for negotiating total compensation at each stage of the hiring process.
Understanding Total Compensation in AI Engineering
The shift toward AI-specialised roles has produced compensation structures that are materially different from traditional software engineering packages. Engineers who evaluate offers purely on base salary are often making decisions with incomplete information that costs them $30,000 to $100,000 in annual total compensation.
At frontier AI companies including Anthropic, OpenAI, and the leading AI labs, base salary is frequently the smallest component of total compensation. A senior AI engineer at a frontier lab might earn $200,000 in base salary, $150,000 in annual RSU vesting, and $30,000 in bonus - a total of $380,000 that looks nothing like the base salary headline. Negotiating from the base salary alone, without understanding the equity structure, vesting schedule, and performance bonus mechanics, is a systematic error that favours employers who have studied this market more carefully than most candidates.
At AI-native startups, equity is typically a more significant component and is structured differently. The percentage of the company, the strike price relative to the current 409A valuation, the vesting schedule, and the liquidation preferences and other cap table terms that determine how much of a liquidity event employees actually receive are all negotiable dimensions that most engineers do not fully explore.
At established enterprises deploying AI (financial services, healthcare, manufacturing), the compensation structure is typically more base-salary-heavy but may include performance bonuses tied to specific AI project outcomes, sign-on bonuses for candidates who reduce LTIP (long-term incentive plan) vesting value from a prior employer, and equity refreshes tied to performance milestones.
When and How to Negotiate
The optimal time to negotiate is after you have a written offer and before you sign. Not before you have an offer (discussing salary expectations before an offer is made typically anchors you lower than you would otherwise land), and not after you have signed (because the leverage disappears the moment you commit).
The single most effective negotiation tactic in the AI engineering market is having a competing offer. The AI engineering talent market is tight enough that most employers will match or beat a competing offer rather than lose a candidate they have invested in screening. Engineers who have one offer should, if feasible, accelerate conversations with other potential employers to produce a competing offer that creates negotiating leverage.
If a competing offer is not available, the second most effective approach is demonstrating specific market data that supports a higher range. Salary data from Built In, Glassdoor, Levels.fyi (the most detailed source for Big Tech compensation), and the recruiter surveys cited in this article provide the factual foundation for a counter-offer request.
The specific request framing that produces the best outcomes is: “I am very interested in this role and excited about the opportunity. Based on my research on the market for [specific skill combination] in [location], and the level of experience and production portfolio I bring, I was expecting a package in the range of [specific range]. Is there flexibility to reach [specific target]?” This framing is specific, ties the request to market data and your specific value, expresses genuine interest, and asks an open question rather than making an ultimatum.
The Equity Conversation
For offers at AI-native companies, the equity conversation is often more important than the salary conversation. The specific questions to ask before accepting any equity-based compensation package:
What is the current common stock valuation (409A) and how recently was it completed? The ratio of strike price to 409A determines the paper value of your options, and a stale 409A from a period of high valuations may not reflect current value.
What is the total fully diluted share count, and what percentage of that does the offered grant represent? Converting share counts to percentages allows meaningful comparison across different-sized companies.
What were the terms of the most recent preferred stock round? Liquidation preferences, anti-dilution provisions, and participating preferred terms all affect how much common shareholders (including employees) receive in a liquidity event. Heavily investor-favourable terms can make employee equity worth far less than it appears.
What is the vesting schedule, and are there any acceleration provisions upon change of control? Standard four-year vesting with a one-year cliff is the norm; understanding any deviations is important.
Has the company raised enough capital to cover at least eighteen months of operations at its current burn rate? Companies with insufficient runway cannot honour multi-year vesting commitments if they run out of money before you vest.
Part Ten: The Network as Infrastructure - Building the Relationships That Actually Produce Opportunities
Job market data consistently shows that 70% to 80% of positions are filled through professional relationships rather than through direct application to posted jobs. This holds in the AI engineering market even more strongly than in the broader market, because the AI engineering community is small enough that reputation and relationships are highly visible and highly influential.
The Three Relationship Tiers That Matter
The relationships most directly relevant to job opportunities fall into three tiers with different investment requirements and different return profiles.
Tier One relationships are former managers and colleagues who can provide referrals, references, and insider information about open roles. These relationships are built over time and maintained through genuine reciprocal engagement - sharing relevant information, congratulating genuine achievements, and staying in contact even when there is no immediate need. Engineers who only contact former colleagues when they need something are not maintaining relationships; they are making withdrawals from accounts they have not maintained.
For every engineer who has worked in technology for five or more years, there should be a list of ten to twenty former managers and colleagues whose good opinion matters and who might be in a position to provide a warm introduction to a hiring manager in a target company. Maintaining those relationships through brief, genuine contact three or four times per year is the minimum maintenance investment that keeps the relationship viable.
Tier Two relationships are practitioners in the target role or target company who can provide information, perspective, and potentially referrals. These are built through community engagement: attending meetups, contributing to open-source projects, engaging with technical content on LinkedIn, and building a public professional presence through writing or speaking.
The AI engineering community, despite its rapid growth, is still small enough that engineers who produce genuinely useful technical content - blog posts that solve real problems, open-source tools that the community uses, conference talks that present real production experience - become recognisable within the community faster than in more established fields. This visibility is directly monetisable in the job market because hiring managers actively look for candidates with community reputation.
Tier Three relationships are weaker ties - LinkedIn connections, Twitter mutual follows, forum participants - who may be irrelevant individually but who collectively provide access to information about opportunities, company cultures, and market conditions that would not otherwise be accessible. The value of maintaining a broad Tier Three network is primarily informational: the news about a company’s aggressive hiring before it becomes public, the information about a team’s culture from someone two steps removed, the awareness that a specific hiring manager at a target company is particularly open to career changers.
LinkedIn as an Active Tool, Not a Passive Presence
Most engineers treat LinkedIn as a resume repository that they update when job searching. Engineers who use LinkedIn as an active professional network tool have materially different job market experiences.
Active LinkedIn use in the context of a job search or career transition involves: publishing short-form technical content about problems you have solved and how you solved them; commenting substantively on technical posts by practitioners in your target role; sending personalised connection requests to people you meet at events or whose content you engage with regularly; and responding to recruiter messages selectively, declining the ones that are clearly wrong, and engaging substantively with the ones that are in the right direction.
The specific content most effective on LinkedIn for AI engineering positioning is: technical tutorials that solve specific, non-trivial problems (how you built a production RAG system, what you learned debugging an MLOps pipeline, how you approached AI security testing for a specific application type); honest retrospectives on what did not work in a project and what you learned; and brief career transition updates that share your progress and invite engagement.
The audience for this content is primarily recruiters and engineering managers who are looking for candidates with the specific skills you are building. The engagement rate on technical content from practitioners is substantially higher than on generic career advice content, and the signal-to-noise ratio for the connections it generates is much higher.
Open Source Contributions as Relationship Capital
Contributing to open-source AI projects is both a skill-building exercise and a network-building exercise. The contributions most visible to potential employers are those in actively maintained, widely used repositories: the Hugging Face datasets and models library, LangChain or LlamaIndex, MLflow, Evidently AI, and the cloud provider SDKs for AI services.
Even small, well-executed contributions - fixing a bug with a clear explanation, adding a missing example to documentation, improving error messages - create a visible record of technical engagement that recruiters and hiring managers can evaluate. They also create opportunities for direct interaction with the maintainers and contributors of those repositories, who are typically practitioners at the companies that hire AI engineers.
The GitHub contribution graph is the first thing many technical interviewers look at after the resume. An active contribution history that shows consistent, recent, AI-relevant work is a more credible signal than a list of bullet points on a resume.
Part Eleven: Industry-Specific Pivots - Where Your Existing Domain Expertise Creates Advantage
The most powerful career strategy for experienced engineers is not abandoning their existing domain expertise to become generic AI engineers. It is combining their domain expertise with AI skills to create a hybrid profile that is rare, highly valued, and difficult for AI to replicate.
Financial Services: The Highest-Compensation Domain Combination
Engineers with experience in financial services technology - payment processing, trading systems, risk management, regulatory reporting, or banking infrastructure - combined with AI engineering skills are among the most sought-after profiles in the current market.
The specific reasons are structural. Financial services companies are deploying AI aggressively for fraud detection, credit decisioning, market analysis, regulatory reporting automation, and customer interaction. The deployment of AI in regulated financial services requires engineers who understand both the AI systems and the regulatory environment: model risk management requirements, fair lending rules, model explainability requirements for credit decisions, and data privacy requirements under applicable regulations.
The FDIC, OCC, and SEC all have guidance on the use of AI in financial services that creates compliance requirements. Engineers who understand both how RAG systems work and how to document those systems for model risk management examination are simultaneously rare and highly compensated.
The salary range for AI engineers with financial services domain expertise at the intersection of AI and compliance is $180,000 to $280,000, with senior roles at quantitative hedge funds and the largest investment banks reaching $400,000 and above.
Healthcare: The Regulatory Complexity Creates Durable Value
Healthcare is the sector where the combination of technical AI skills and domain expertise creates the most durable competitive advantage, because the regulatory complexity of healthcare AI deployment creates barriers to entry that pure AI engineers cannot overcome without domain knowledge.
HIPAA compliance for AI systems, FDA guidance on software as a medical device, CMS requirements for AI tools used in clinical documentation, and the clinical validation requirements for AI systems that influence patient care are all domains where technical AI skills without healthcare domain knowledge are insufficient for responsible deployment.
Engineers with healthcare IT experience who develop AI engineering skills are positioned for roles in health technology companies, hospital system IT departments, pharmaceutical companies building AI tools for drug discovery or clinical trials, and the growing number of AI health startups that require both dimensions.
Compensation for healthcare AI engineers with domain expertise ranges from $150,000 to $230,000 in most markets, with the top end associated with senior roles at well-funded health AI companies or the largest health systems.
Manufacturing and Industrial: The Least Competitive, Most Growing Opportunity
Of all the domain-AI combinations, manufacturing and industrial AI may be the most underserved and fastest growing. The combination of AI tools with operational technology (OT), industrial IoT, computer vision for quality control, and predictive maintenance is creating an engineering market that is growing rapidly but attracting fewer candidates than healthcare or financial services AI because manufacturing is less prestigious in the engineering community than consumer or financial technology.
Engineers with backgrounds in automation, embedded systems, manufacturing execution systems, industrial control systems, or supply chain technology who add AI engineering skills are positioned for one of the fastest-growing markets in industrial AI with some of the most direct competitive advantages. The companies building AI for manufacturing - including large industrial companies like GE, Siemens, and Honeywell, and AI-native startups applying AI to industrial problems - are actively seeking engineers who understand both domains.
The salary range is $140,000 to $220,000 for experienced engineers at the domain-AI intersection in manufacturing, with significant upside from equity at early-stage industrial AI startups.
Legal Technology: The Compliance and Reasoning Intersection
Legal technology is undergoing significant AI disruption, but the disruption is creating a wave of hiring for engineers who can build AI tools for legal applications that comply with the ethical and reliability requirements of the legal profession. E-discovery platforms, contract analysis tools, legal research tools, and regulatory compliance automation are all areas of active investment and hiring.
The legal services domain is unusual in that the standards for accuracy, reliability, and explainability are among the highest of any professional domain. A financial AI tool that makes an error costs money. A legal AI tool that makes an error in a litigation context or a compliance decision can have consequences that far exceed the financial cost. This creates demand for AI engineers who understand not just how to build AI systems but how to validate them to the reliability standards that legal applications require.
Engineers with experience in compliance technology, e-discovery, legal data management, or adjacent fields who develop AI skills are positioned for a market where the combination is genuinely hard to find. Compensation ranges from $150,000 to $220,000 at legal technology companies and law firms with technology development capabilities.
Education Technology: The Long Game with Social Impact
Education technology has been one of the most disrupted sectors in the current AI wave - Chegg’s 45% workforce reduction, driven by students abandoning homework help services for AI tools, is the most visible example. However, the disruption creates a rebuilding opportunity for engineers who want to build the next generation of AI-enhanced educational technology.
The specific problems worth solving in education AI are those where AI complements rather than replaces the teacher-student relationship: adaptive learning systems that personalise curricula to individual students, tools that help teachers identify struggling students early, AI tutoring systems that provide feedback at the level of detail and patience that human tutors cannot sustainably provide, and assessment tools that evaluate understanding rather than test-taking compliance.
The compensation for education AI roles is lower than most other AI specialisations - typically $120,000 to $175,000 - but engineers who prioritise social impact alongside compensation find this market compelling. The venture capital investment flowing into AI education companies is substantial, and early engineers at successful AI education companies can capture significant equity upside.
A Closing Note on the Citizenship Advantage in Context
Throughout this article, the citizenship advantage has been framed primarily as an employment access issue - the clearances, government positions, and defence technology roles that require citizenship. It is worth briefly broadening that framing to note that citizenship also provides a form of career resilience that is less visible but equally important.
The H-1B visa holders who are among the hardest hit in the current layoff wave face a compounding problem that US citizens do not: the 60-day clock on finding new employment, the immigration processing uncertainty, and the geographic constraints of visa sponsorship all add friction to career transitions that US citizens navigate without. The US citizen engineer who is laid off is managing a difficult professional and financial situation. The H-1B holder in the same situation is managing those challenges plus the possibility of forced departure from a country where they have built their lives.
This is not a reason for citizens to be less concerned about the employment disruption - the displacement is real for everyone. It is a reminder that the ability to take the time needed to retrain effectively, to consider the full range of career options including government work and defence contracting, and to negotiate from a position of geographic flexibility rather than geographic constraint is a meaningful structural advantage that should inform how aggressively and how deliberately it is used.
The engineers who navigate this transition best will be those who understand both the threats and the advantages with the same honest clarity, and who act on that understanding with the consistency and urgency the moment requires.
Published March 25, 2026 by the InsightCrunch Research Team. Salary data sourced from IEEE-USA, Built In, Glassdoor, Robert Half 2026 Technology Salary Report, and Axiom Recruit. Hiring trend data from SSi People, SignalFire, NACE Job Outlook 2026, and BLS occupational projections. Layoff data from Layoffs.fyi, TrueUp.io, and Crunchbase. Career guidance based on practitioner accounts, recruiter interviews, and published labour market research. All salary figures in USD for US-based positions unless noted.
Part Twelve: The Interview Process for AI Engineering Roles - What Has Changed
The interview process for AI engineering roles in 2026 is materially different from the traditional software engineering interview process, and engineers who prepare using outdated frameworks will underperform their actual competence level. This section describes what employers in the AI engineering space are actually testing for and how to prepare effectively.
The Technical Interview Has Changed
The classic FAANG-style coding interview - algorithm problems on a whiteboard or in a code editor, typically drawn from LeetCode-style problems - is being modified or abandoned by many AI engineering employers. There are two reasons. The first is that Claude Code and similar tools have made rote algorithm recall less relevant as a proxy for engineering quality. The second is that AI engineering interviews need to test different competencies than traditional software engineering interviews.
The interview formats most common in AI engineering hiring in 2026 include:
System design interviews that extend into AI system design: not just “design a URL shortener” but “design a RAG system for a financial services company that needs to answer questions about regulatory filings” or “design a monitoring system for an LLM-powered customer service agent.” These interviews test whether candidates understand the specific design challenges of AI systems - vector database selection and indexing, embedding quality, retrieval evaluation, hallucination detection, latency management - alongside the general distributed systems knowledge that system design interviews have always tested.
Portfolio review interviews where candidates walk through a project they have built, explaining the decisions they made, what did not work, and what they would do differently. These interviews heavily favour candidates who have built real AI systems and can speak to production challenges rather than candidates who have only done tutorial projects.
AI tool proficiency demonstrations where candidates are asked to solve a problem using AI coding tools in real time. These are live sessions where candidates use Claude Code, GitHub Copilot, or equivalent tools to build something under observation. They test whether candidates are genuinely proficient with these tools or just claim to be - a distinction that is increasingly important as proficiency becomes a standard expectation.
Take-home technical assessments that involve building a small but complete AI system within a defined scope. These are given because they better reflect the actual work of the role than coding puzzles, and they allow employers to evaluate code quality, documentation habits, and decision-making in a realistic context.
The Behavioural Interview in the AI Era
Behavioural interviews for AI engineering roles test competencies that are different from those tested in traditional software engineering roles, reflecting the different challenges of working with AI systems.
Failure and learning: AI systems fail in ways that traditional software systems do not - hallucinations, distribution shift, adversarial inputs, degradation over time. Interviewers want to understand how candidates identify, diagnose, and learn from these failures, because the ability to reason about AI failure modes is not something that traditional software engineering experience necessarily develops.
Ethical reasoning: How do you think about fairness, bias, and the potential harms of AI systems you build? What would you do if you discovered that a system you built was producing discriminatory outputs? These questions are increasingly standard in AI engineering interviews at companies that are serious about responsible AI deployment.
Learning agility: The AI field moves faster than any other area of software engineering. Interviewers want to understand how candidates stay current, what their learning process looks like, and how they approach a new problem domain. The answer “I read papers and build projects” is more credible when it is accompanied by specific recent examples than when it is a general claim.
What Distinguishes Good Candidates from Great Ones
After reviewing the patterns in AI engineering hiring based on practitioner accounts and recruiter interviews, the distinguishing characteristics of candidates who consistently succeed in AI engineering interviews are:
They can explain what they did and why, not just that they did it. The candidate who says “I used Pinecone as the vector database” is less impressive than the candidate who says “I chose Pinecone over Chroma for this use case because the production scale required managed infrastructure, and I evaluated both on retrieval latency and cost before making the decision.” The reasoning process demonstrates genuine production thinking rather than tutorial reproduction.
They are honest about what they do not know. AI engineering is a new enough field that no candidate knows everything, and interviewers know this. Candidates who pretend to know things they do not, or who give confident wrong answers, create more concern than candidates who say “I have not worked with that specific tool, but here is how I would approach learning it.”
They have opinions about trade-offs. The strongest AI engineering candidates have developed views about when different approaches make sense and are willing to defend those views with evidence. The candidate who always says “it depends” without following up with the specific factors that determine the decision is signalling that they have not yet built the judgment that comes from production experience.
They can talk about evaluation. AI systems are only as good as your ability to measure whether they are working. Candidates who can speak specifically about how they measured the quality of AI outputs - what metrics they used, how they designed evaluation datasets, how they detected degradation over time - are demonstrating the practical rigor that distinguishes engineers who have shipped AI systems from those who have only built them.
Part Thirteen: The 2028 Horizon - Planning Beyond the Immediate Transition
The career decisions of 2026 have implications that extend well beyond the immediate transition period. Positioning for the three to five year horizon requires thinking beyond the current market conditions to the state of the AI employment market as capability continues to advance and enterprise adoption continues to mature.
The Likely State of the Market in 2028
Based on the capability trajectory described in the earlier articles in this series, and on the historical pattern of enterprise technology adoption cycles, the AI engineering job market in 2028 is likely to have the following characteristics:
Production MLOps will be more standardised. The tools for deploying, monitoring, and maintaining AI models in production will be more mature, more standardised, and easier to use than they are in 2026. This means the differentiation value of MLOps skills will decline as the tools commoditise, similar to how DevOps skills declined in value as platforms like GitHub Actions and Kubernetes matured and became standard. Engineers who build MLOps skills in 2026 and 2027 are capturing the highest premium, but should plan to layer on deeper specialisation as the baseline commoditises.
AI safety and governance will be a larger, more structured field. The regulatory environment for AI in enterprise applications will be more developed by 2028, with clearer requirements and established frameworks. The engineers who build expertise in AI governance and compliance in 2026 and 2027 will be ahead of a market that will be substantially larger by 2028.
Agentic AI systems will be deployed at much wider scale. The Agent Teams capability of current Claude models, and the competing agentic frameworks from OpenAI and others, will produce a wave of enterprise agentic deployments that is just beginning in 2026. The engineering discipline of building, validating, and maintaining agentic AI systems - systems that autonomously take actions in the world rather than just generating text - will be one of the most valuable specialisations in the 2027-2030 period.
Domain-specific AI models will be more common. The trend toward fine-tuning foundation models for specific domain applications - healthcare AI models, legal AI models, financial AI models - will continue and accelerate. Engineers who have built experience with fine-tuning and with domain-specific model evaluation will have the skills most relevant to this deployment pattern.
Skills to Build Now for the 2028 Market
Engineers who are planning beyond the immediate transition should consider investing in the following skills that are early in the adoption curve in 2026 but will be at the peak of their market value in 2027 to 2029:
Agentic system design and safety: Understanding how to build AI agents that are reliable, safe, and auditable in production settings. This includes both the technical dimension (designing agentic architectures, managing agent state, preventing unintended actions) and the governance dimension (logging and monitoring agent behaviour, designing human oversight checkpoints, creating evaluation frameworks for agentic systems).
Multimodal AI integration: Systems that process text, images, audio, and video together are becoming more common in enterprise applications. The engineering challenges of multimodal AI systems - managing different modalities’ data pipelines, evaluating multimodal outputs, handling modality-specific failure modes - will be a growing specialisation.
AI hardware and inference optimisation: As AI inference costs become a larger share of enterprise AI budgets, the engineering discipline of optimising how models run - through quantisation, distillation, hardware selection, and inference architecture design - will grow in value. This is a more specialised area that requires deeper ML knowledge than most application engineering roles, but the compensation ceiling is high.
Cross-company AI governance and standards work: The engineers who participate in industry standards work - through IEEE AI standards committees, NIST AI framework development, or sector-specific standards bodies in healthcare and financial services - are building credibility and relationships that will be particularly valuable as the regulatory landscape matures.
The Entrepreneurship Window
The period from 2026 to 2028 may be the most favourable window for AI-native entrepreneurship that will exist in the near term. The combination of increasingly capable AI tools that allow small teams to build products that previously required large teams, deep domain expertise from years of enterprise experience, and venture capital appetite for AI applications creates conditions where experienced engineers with domain knowledge can build genuinely competitive companies.
The window may narrow as the market matures. Competition in AI applications will intensify as more entrepreneurs enter the space and as the largest AI companies extend their product platforms to cover more use cases. Engineers who are considering entrepreneurship should evaluate whether the current moment, with its specific combination of accessible frontier AI capabilities, domain expertise, and capital availability, represents a better moment to act than the hypothetical future moment when conditions might seem safer.
The answer is highly individual. Engineers with deep domain expertise, strong customer insight in a specific vertical, and the personal financial cushion to tolerate a startup’s risk profile should seriously evaluate the next twelve to eighteen months as a potentially optimal founding window. Engineers without those ingredients should focus on the employment strategies described in earlier sections.
The Final Honest Assessment
This article has covered a lot of ground, from the specific functions being automated to the certification pathways worth pursuing, from the salary data for each AI engineering specialty to the mental health management strategies for the transition period. The goal throughout has been specificity rather than generality, because the career decisions of 2026 require specific, evidence-based guidance rather than broad reassurances or broad warnings.
The final honest assessment is this: US citizen engineers who engage actively with the transition - who invest in the right skills, who use their citizenship advantages deliberately, who build their networks before they need them, and who manage the psychological challenges of a difficult period with the same intentionality they bring to technical problems - have genuinely good career prospects in the AI era.
The transition is real and the disruption is serious. The salary premium for AI engineering skills is also real and substantial. The clearance pathway is genuinely accessible for the majority of US citizens and provides employment stability that the commercial sector cannot currently match. The government and defence technology sector is growing while commercial Big Tech is cutting. The domain expertise that experienced engineers have built over ten or fifteen years is more valuable than ever when combined with AI skills rather than replaced by AI.
None of this makes the transition easy. It makes it navigable with clear sight of where to go and honest effort in getting there. That is the most useful thing a career article can provide in a market this uncertain: not false confidence, not paralyzing fear, but an accurate map and specific directions for the next leg of the journey.
The territory is changing. The map is here. The movement is yours.
This is the third article in InsightCrunch’s series on the IT sector layoff wave of 2026. The first article covered the twenty largest global IT employers and their specific layoff numbers. The second covered Anthropic’s specific role in the disruption. This article provides actionable guidance for US citizen engineers navigating the transition. All three articles are available in the Industry category.
Part Fourteen: The Resume and LinkedIn Profile for the AI Era
The mechanics of how US engineers present themselves in the AI era job market require specific attention because the conventions have shifted alongside the hiring criteria. This section covers the specific resume and LinkedIn profile optimisations that have the most impact in the current market.
Resume Optimisations That Matter in 2026
The resume for AI engineering roles in 2026 has a different emphasis than the resume for traditional software engineering roles. The five most important changes:
Lead with AI tool proficiency at the top. The traditional resume structure places education at the top for new graduates and professional summary at the top for experienced engineers. For AI engineering job searches in 2026, the most effective structure places a “Technical Skills” or “Core Competencies” section at the top that immediately communicates AI tool proficiency. Hiring managers and recruiting systems are screening for specific AI tools and frameworks (Claude Code, LangChain, MLflow, Pinecone, PyTorch, HuggingFace Transformers, AWS Bedrock, etc.) and a resume that does not make these visible within the first screen is at risk of being screened out before the human reviewer sees it.
Quantify AI project outcomes specifically. The generic bullet point “Led AI project to improve customer service” is dramatically less effective than “Built RAG-based customer inquiry system using Claude API and Pinecone that reduced average handle time by 34% and handled 60,000 monthly queries with 94% user satisfaction, replacing a team of 12 human agents.” The specific numbers, tools, and outcomes give screeners and interviewers concrete evidence to evaluate and make the AI work credible in a way that generic descriptions cannot.
Document GitHub contributions. If you have made meaningful contributions to open-source AI projects, include the specific repositories and the nature of the contributions. “Contributed 23 pull requests to the LlamaIndex documentation and examples library” is verifiable, specific, and creates a credibility signal that is difficult to fake.
Address the AI generation elephant in the room. Many engineers wonder whether to disclose that they used AI tools to generate portions of their resume. The answer is: use AI tools to help draft and improve your resume, but ensure the final product accurately reflects your own experience and voice. A resume that describes experiences you do not have, or that claims skills you cannot demonstrate, creates risks in interviews that are not worth the screening advantage. Using AI to improve the clarity and structure of genuinely accurate descriptions of your own work is appropriate and increasingly standard.
Include a brief technical projects section with links. A resume section listing three to five AI projects with one-sentence descriptions and GitHub or demo links is worth more than an equivalent amount of space describing job responsibilities in conventional bullet points. The projects section is what hiring managers look at most carefully after the work experience section for AI engineering candidates.
The LinkedIn Profile Optimisation Checklist
The LinkedIn profile elements most directly linked to recruiter outreach and hiring manager attention in the AI engineering market:
| Headline: Include your AI specialisation explicitly in your headline, not just your job title. “MLOps Engineer | Production AI Deployment | AWS Bedrock | LangChain” is significantly more searchable and specific than “Senior Software Engineer at [Company Name].” |
About section: Lead with your AI engineering philosophy and the problems you love solving, not with a summary of your work history. The about section is where you establish voice and differentiation in a way that a resume does not allow. Include the specific AI frameworks and tools you use most frequently, the types of problems you have solved, and what you are looking for in your next role.
Featured section: Pin your three best portfolio pieces - GitHub repositories, technical blog posts, demo videos of projects you have built - to your featured section. This is the most visited section of your profile after the headline and about section, and it is your best opportunity to show rather than tell what you can do.
Skills section: Update your skills to reflect AI-specific competencies prominently. LinkedIn’s recruiter search tools filter heavily on skills, and ensuring that your profile includes the specific technical skills that match the roles you are targeting is basic search engine optimisation for your professional profile.
Recommendations: Request updated recommendations specifically from former colleagues or managers who worked with you on AI projects. Recommendations from 2019 that praise your Java skills are less valuable than a 2025 recommendation that describes your specific AI engineering work.
Creator mode: Enabling LinkedIn’s Creator mode and publishing AI engineering content - even brief observations about a problem you solved, a tool you tried, a decision you made and why - creates the kind of visible technical engagement that differentiates active practitioners from passive profile maintainers. The engineers with the most recruiter outreach are those who combine current AI skills on their profile with visible technical engagement through their content activity.
Part Fifteen: The Support Ecosystem - Resources and Communities Worth Your Time
The quality of the support ecosystem available to US engineers navigating the current transition is substantially better than the ecosystem available in previous technology transitions, because the AI engineering community has built documentation, educational resources, and professional communities faster than any prior specialisation has. This section catalogues the highest-value resources across categories.
Technical Communities and Forums
The Hugging Face community forum and Discord server are the best single destinations for practitioners working with transformer models, fine-tuning, and the broader open-source ML ecosystem. The quality of discussion is high, the participants are active practitioners rather than students, and questions about specific production challenges get substantive responses from people who have solved the same problems.
The MLOps Community Slack is the most active practitioner community for ML operations, model deployment, and production ML topics. The channels covering monitoring, model serving, and MLOps platform comparisons are particularly useful for engineers building in these areas.
The MLIR (Machine Learning Infrastructure) community, while more research-oriented, is valuable for engineers working on hardware optimisation and inference efficiency. The documentation and discussion in this community are at the frontier of what is possible in model deployment efficiency.
For cleared technology careers, ClearedJobs.Net and ClearanceJobs provide community forums and job listings specific to the cleared technology sector. The forums include practical advice about the clearance application process from practitioners who have navigated it.
Career Transition Support
Several organisations provide specific support for technology workers navigating career transitions that are worth knowing:
The Tech Career Compass, a nonprofit that provides career coaching and community support for technology workers who have been laid off, has grown substantially during the current cycle. Their programmes include group coaching, mock interviews, and a mentorship network.
Code2College, Creating Coding Careers, and similar organisations that support career transitions into and within technology provide structured programmes that some mid-career engineers find valuable for accountability and support during retraining periods.
Hackathons and AI competitions (Kaggle competitions for ML, various LLM hackathons) provide structured skill-building exercises with the added benefit of public visibility through leaderboard placement and submission portfolios. Kaggle in particular provides a structured competitive environment where engineers can demonstrate specific AI skills with results that are independently verifiable.
Financial Support Resources
For engineers who are between jobs and managing the financial transition, several resources beyond unemployment insurance are worth knowing:
The Workforce Innovation and Opportunity Act (WIOA) provides federally funded retraining support for workers who have been displaced from their jobs. State workforce development agencies administer WIOA funding, and some of the funded programmes include technology retraining relevant to AI engineering transitions.
Employee stock purchase plans and 401k funds become relevant during gap periods in ways that they are not during continuous employment. Understanding the rules for withdrawals, the COBRA continuation options for health insurance, and the tax implications of various actions during a gap period is worth the time investment with a fee-only financial advisor.
For engineers who were H-1B workers on a path to permanent residency and who recently received citizenship, the transition to citizen status means that previously available employment options in the cleared sector are now accessible. The timing of this transition relative to the current job market creates specific opportunities that may not exist in future cycles when the cleared sector market may be more competitive.
Three-part series on InsightCrunch covering the 2026 IT sector layoff wave. Article one: The twenty largest global IT employers and their specific layoff counts. Article two: Anthropic’s specific role in accelerating the employment disruption. Article three: This article - what US citizen engineers can actually do. Each article is approximately 25,000 words with comprehensive FAQ sections and cited, verified data throughout.
Part Sixteen: Seven Specific Career Profiles - What Each Type of Engineer Should Do
The preceding sections provide broad strategic guidance that applies across different engineering career profiles. This section applies that guidance to seven specific engineering profiles, with the specificity that the abstract guidance cannot provide. Find the profile that most closely matches your situation and treat that section as your personal action plan.
Profile One: The Laid-Off Amazon Software Development Engineer, Six Years Experience, Seattle
Situation: You were part of Amazon’s January 2026 restructuring. You have six years of experience primarily in Java backend development, API development, and AWS service integration. Your NQT and system design interviews are solid. You have no specific AI skills yet. Your severance covers six months.
Immediate priority: File for unemployment insurance on day one. Update your LinkedIn to “open to work” using the private setting (visible only to recruiters). Do not begin bulk-applying to SDE roles at other large tech companies - that market is saturated with Amazon alumni from multiple recent rounds.
Skill investment: Your existing AWS knowledge is your most valuable transitional asset. The fastest path to a marketable premium position from your current skill set is: AWS ML Specialty certification combined with a RAG project that uses AWS Bedrock and RDS with pgvector (keeping the stack native AWS, which plays to what you already know). Timeline to a marketable credential plus portfolio project: three to four months of concentrated effort.
Job search targets: AWS-focused AI engineering roles at mid-size enterprise companies, Amazon vendor partners (companies that sell into AWS), and AWS consulting partners who are hiring AI deployment engineers. The Tier 2 tech companies that compete with Amazon’s cloud customers - the Fortune 500 companies building on AWS - are a large hiring pool that is less saturated with Amazon alumni than the pure tech employer pool.
Salary target: $165,000 to $190,000 is the realistic range for your experience level and the skills you will have after three to four months of focused investment. Clearing the AWS ML Specialty and having a RAG project on GitHub is the minimum portfolio for that range.
Profile Two: The Senior Google/Alphabet Engineer, Fifteen Years, Bay Area, Android Team
Situation: You were in Google’s Android team restructuring. Fifteen years of experience primarily in C++, Java, and Kotlin for Android platform work. You have spent your entire career at Google. The Bay Area is home but your RSU vesting is complete.
Immediate priority: Do not rush. Your financial position with fifteen years of Google compensation and complete RSU vesting is strong enough to do this transition carefully. Resist the pressure to take the first acceptable offer. Use the first thirty days for assessment, not applications.
Skill investment: Your strongest card is systems-level Android and mobile platform knowledge combined with Google’s internal engineering culture and scale experience. The intersection most worth pursuing: AI on-device systems (edge AI, LLM inference optimisation for mobile, on-device model deployment). Google, Apple, and the growing ecosystem of device manufacturers are building AI capabilities that run on device rather than in the cloud, and the expertise to do this well is extremely rare.
Alternative if edge AI does not appeal: Your fifteen years of C++ and systems work creates a direct path into AI infrastructure engineering. The companies building inference engines (Nvidia’s TensorRT, custom inference solutions for specific hardware) need engineers who understand systems-level optimisation and who can work in C++ on performance-critical code. This is closer to your existing skill set than most AI engineering pivots.
Job search targets: Apple Silicon AI team, Qualcomm AI research, Nvidia AI platform team for mobile inference, and the device AI teams at Samsung, Meta (Reality Labs hardware), and a growing set of AI hardware startups. These are positions that the general Android developer population cannot target because they require the specific combination of Android platform expertise and inference optimisation interest.
Salary target: $250,000 to $400,000 in total compensation is realistic for your experience level in the target roles, particularly at Apple and Nvidia where competition for senior Android platform expertise is intense.
Profile Three: The Mid-Level QA Engineer, Seven Years, Austin, Pivoting
Situation: Your role at a mid-size SaaS company has been formally announced as a target for “automation” in the company’s AI strategy. You have seven years of manual QA experience, some basic Python for test scripting, and you are watching colleagues transition out. You have not been laid off yet but the writing is clear.
Immediate priority: Do not wait for the layoff announcement to begin your transition. Use the remaining employed time - and your employer’s training budget if available - to begin the skill investment now. Initiating a transition while employed is faster, lower-stress, and gives you the ability to be selective about your next role.
Skill investment: The targeted pivot is test automation engineer with an AI testing specialisation. The specific trajectory: move from manual test execution to AI-assisted test automation framework design. The tools to learn are: Playwright or Selenium with AI-assisted test generation extensions, the pytest framework for Python-based test automation, and GitHub Actions for CI/CD integration. The AI-specific addition is learning to work with Anthropic’s or OpenAI’s APIs to build intelligent test case generation systems that identify edge cases manual testing misses.
The more ambitious pivot if you want a compensation step-change: AI QA engineer at an AI product company. These roles are responsible for evaluating AI output quality, designing evaluation datasets for LLM products, and building the measurement systems that tell a company whether its AI is working. This combines your QA domain knowledge with the AI-specific dimension and commands $120,000 to $160,000 versus the $85,000 to $100,000 range for traditional QA.
Job search targets: AI product companies (the ecosystem of companies building on Claude, GPT-4, and Gemini) that need engineers to evaluate AI output quality. QA automation roles at companies adopting AI development workflows where your understanding of both manual and automated testing creates a bridging role that is valuable during the transition. Defence and government contractors with AI testing requirements (citizenship advantage applies here).
Profile Four: The Federal Government Contractor Engineer, Three Years, DC Area
Situation: You have three years of experience at a medium-size government IT contractor in Northern Virginia. Your work has been primarily SharePoint administration, basic Python scripting, and IT project coordination. Your work is Secret-cleared. You are not at immediate risk but you want to move up.
Immediate priority: Your secret clearance is a valuable career asset that most engineers do not have. You are already inside the cleared ecosystem. The question is how to leverage it into roles with higher compensation and more interesting technical work.
Skill investment: The highest-ROI path from your specific starting point is the intersection of your clearance and AI capabilities for government applications. The federal government’s AI strategy, executed through agencies including the DoD, DHS, and civilian agencies, is creating demand for engineers who can deploy and govern AI tools within the security constraints of government systems. Learning to work with Claude Gov, deploying AI tools on government cloud infrastructure (AWS GovCloud, Microsoft Azure Government), and understanding FedRAMP AI compliance requirements positions you for roles at the intersection of your existing cleared ecosystem experience and the growing government AI market.
Job search targets: Move from your current mid-size contractor to one of the Big Four cleared technology contractors (Booz Allen Hamilton, Leidos, SAIC, Peraton) or to a pure-play government AI company. The latter are growing rapidly and often provide better compensation and more interesting technical work than the legacy contractors.
Salary target: $120,000 to $150,000 in your next role is realistic based on your cleared status and the AI skills you will add. Within three to five years of continued investment, $170,000 to $200,000 is achievable.
Profile Five: The Freelance/Contract Engineer, Intermittently Employed, Multiple Short Stints
Situation: You have been freelancing and contracting for the last five years after leaving a full-time role. You have strong Python skills, have done data engineering and basic ML work for various clients, but do not have a continuous employment narrative that reads well on a resume.
Immediate priority: Consolidate the strongest of your contract work into a coherent professional narrative. The resume needs to frame your freelance period in terms of the outcomes you produced for clients, not the list of short engagements. “Built and deployed three production ML systems for enterprise clients in financial services and healthcare, managing full lifecycle from data pipeline to model serving and monitoring” is more effective than a list of three to six month contracts.
Skill investment: Your project-based background is actually advantageous for AI engineering, because AI engineering is fundamentally project-oriented. Build three to five strong portfolio projects that demonstrate the specific AI skills in demand - RAG, MLOps, AI security - and document them thoroughly on GitHub. The portfolio substitutes for the continuous employment narrative that is harder to construct from a freelance background.
Job search targets: AI-native startups (which have less bias against non-linear career histories than large companies) and contract-to-hire roles at companies that are building AI teams. Your existing freelance relationships may be your most valuable job search asset - former clients who trust your work are often willing to provide referrals or to convert a relationship into a more structured engagement.
Salary target: The freelance premium (typically 20% to 30% above equivalent full-time salary for contract work) means your day rate calculation should be: target full-time equivalent salary / 52 weeks / 5 days = daily rate target, then add the freelance premium. For AI engineering skills at your experience level, the target full-time equivalent is $140,000 to $165,000, which translates to a day rate of $700 to $900.
Profile Six: The Recent Computer Science Graduate, No Industry Experience
Situation: You graduated in May 2025 with a CS degree. You have been applying to traditional software engineering jobs for ten months with limited success. Entry-level SDE hiring at large tech companies is down 25% from its 2023 peak. You have university projects but no industry experience.
Immediate priority: Accept that the traditional path (apply to big tech, get SWE offer, start with a base of $120,000-$150,000) is significantly harder in 2026 than it was when you were deciding to major in CS. This does not mean your degree was wrong; it means the market has changed and your strategy needs to change with it.
The fastest path from your starting point to employed in AI engineering: Government technology entry-level programmes. USDS (US Digital Service), the Technology Modernization Fund, and state government digital service offices actively recruit recent graduates for technology roles. These are lower-pay than Big Tech at entry level but provide real-world experience and the government service track that leads to PSLF for student loans.
Alternatively, AI startup roles: Many well-funded AI startups specifically want recent graduates who are genuinely AI-native - who have been building with AI tools throughout their education - over more experienced engineers who are learning AI as a second language. Your competitive advantage is that you have never learned to code without AI assistance, which is increasingly the expectation rather than the exception.
Skill investment: Build two to three AI projects that are demonstrably useful rather than demonstrably clever. The problem with many computer science projects is that they showcase algorithmic sophistication rather than product thinking. Build something that a non-engineer would actually use: a RAG-based tool that helps someone in a specific domain (legal, medical, financial) answer a specific category of questions. Document it, deploy it publicly, and gather user feedback if possible.
Salary target: $90,000 to $120,000 for your first role is a realistic range in the current market for recent graduates with strong AI project portfolios. This is lower than the 2022 peak of Big Tech entry-level offers but is competitive for the market conditions of 2026.
Profile Seven: The Engineer Returning from Caregiving Break, Five to Ten Years Prior Experience
Situation: You left the workforce three to five years ago for caregiving responsibilities. Your technical skills are from the pre-ChatGPT era. You are ready to return but the market has changed dramatically since you left.
This profile requires the most significant technical updating, but it also has a specific advantage: your prior experience means that you are not learning engineering from scratch. You are updating a foundation rather than building from zero, which is faster and more credible to employers.
Immediate priority: Assess your pre-gap skills honestly against current market demand. If your prior skills were in areas that are still foundational (Python, SQL, system design, basic cloud), you have a strong foundation to update. If your prior skills were in areas that have declined in relevance (certain legacy tools, specific frameworks that have been superseded), accept that those skills need to be replaced rather than updated.
Skill investment: The reentry portfolio project strategy is your most important tool. Build one substantial AI project that demonstrates current AI engineering skills and that you can speak to in depth in interviews. The project serves two purposes: it updates your skills and it gives you something recent and relevant to discuss rather than having the entire interview focus on your pre-gap experience.
Job search targets: Companies with explicit returnship programmes (Amazon, Microsoft, Google, and many others have programmes specifically designed for experienced engineers returning from career gaps) are worth exploring. Government and cleared sector roles, which have longer and more thorough evaluation processes that give more weight to demonstrated competence over continuous employment history, are often more accessible for returners than fast-moving startup or Big Tech roles.
Salary target: Returning engineers should not expect to re-enter at the salary level they were at when they left, particularly if they left at a senior level. The market and skills have both changed. A realistic re-entry level is one to two years below where you were when you left, with the expectation of returning to your prior compensation level within eighteen to twenty-four months of consistent performance.
This completes the three-part InsightCrunch series on the 2026 IT sector layoff wave. The full series provides the most comprehensive available analysis of the current disruption from the global macro picture, through the specific role of Anthropic, to the actionable career guidance for US citizens navigating it.
Supplemental: The Veteran Engineer Pathway - Military to Tech and Tech to Cleared
One significantly underserved population in discussions of the current IT employment transition is the veteran engineer - either veterans of the US military who are transitioning into civilian technology careers, or civilian engineers who are exploring the military and defence technology sector for the first time. Both populations have specific advantages and specific challenges in the current market.
Military Veterans Transitioning Into Civilian AI Engineering
US military veterans transitioning into civilian technology careers in 2026 are entering one of the most hospitable environments that market has offered in years, for several specific reasons.
Veterans who served in roles involving intelligence analysis, signals intelligence, or cyber operations have directly transferable skills in AI-adjacent disciplines. The pattern recognition, anomaly detection, and adversarial thinking that characterises intelligence and cyber work are directly applicable to AI security engineering, AI evaluation, and the kind of rigorous thinking about failure modes that AI systems require. Veterans from these backgrounds have often done, in a classified context, work that is closely analogous to what civilian AI security engineers are paid to do.
Veterans with active security clearances - particularly TS or TS/SCI - possess the single most valuable credential for the cleared technology sector, and they can transition into defence AI contractor roles with shorter processing delays than civilian applicants. The combination of a cleared status, military work experience in relevant disciplines, and a GI Bill education benefit that covers technical retraining creates a pathway into AI engineering that has unusually strong economics.
The specific retraining investment most valuable for veterans making this transition: Python programming (via the Codecademy, freeCodeCamp, or CS50x pathways), cloud fundamentals (AWS Cloud Practitioner as an entry credential), and then the AI engineering skills described in the Tier 1 section of this article. The full pathway from basic Python to job-marketable AI engineering takes eighteen to twenty-four months for most veterans, which is compatible with a full GI Bill benefit window.
The Tech-to-Cleared Transition at Different Career Stages
For civilian engineers at different career stages who are considering the cleared transition for the first time, the strategic calculus differs:
Early career: The cleared sector at the entry level is accessible through specific STEM-focused federal internship programmes (the DoD SMART scholarship, the Intelligence Community Centres of Academic Excellence programmes) and through entry-level positions at defence contractors that sponsor clearances. The entry-level cleared salary premium over comparable uncleared positions is smaller than at senior levels, but the investment in building a clearance and a reputation within the cleared ecosystem compounds over time.
Mid-career: The cleared transition at the five to fifteen year experience level typically involves applying to cleared contractor positions that match your existing technical skills. The companies most willing to sponsor clearances for mid-career engineers without prior cleared experience are those with significant ongoing hiring needs and long-term contract backlogs - Leidos, Booz Allen Hamilton, and the mid-size defence AI companies are the best targets.
Senior level: For senior engineers considering a cleared transition, direct engagement with the intelligence community technology community through events including the annual Intelligence Community Industry Day, the DoD AI and Data Forum, and similar government technology events provides visibility into the specific skill gaps that drive the most urgent cleared hiring needs. Senior engineers who can speak credibly to the specific technical challenges of government AI deployment are in a position to negotiate directly for senior roles rather than entering through standard application processes.
Published March 25, 2026 by the InsightCrunch Research Team. This article is the third in a three-part series on the 2026 IT sector layoff wave. Feedback on the content, additional career profiles worth addressing, or corrections to specific data points can be submitted through the InsightCrunch contact form.
The Numbers Summarised: Quick Reference Salary Table
For readers who want to return to the compensation data without reading through the full article, this table consolidates the key salary figures cited throughout:
AI and ML Engineering (US, 2026)
AI Engineer average total compensation: $211,243 (Built In). LLM fine-tuning specialist average: $174,727, top performers above $300,000. MLOps engineer average: $165,000; senior roles at AI-first companies $200,000 to $300,000. Data engineer for AI initiatives: $153,750 base, total compensation frequently above $185,000. Cloud AI architect: $135,000 to $175,000 base, total compensation above $200,000 for senior. AI security engineer: $150,000 to $240,000 depending on cleared status. Developer relations for AI platforms: $140,000 to $200,000 plus equity.
By Geography (AI Engineer base, 2026)
San Francisco Bay Area: $212,859 average (Glassdoor). Seattle: approximately $190,000 to $200,000. New York City: approximately $185,000 to $195,000. Austin, Denver, other mid-tier markets: approximately $155,000 to $170,000. Northern Virginia (cleared sector): $165,000 to $200,000 with cleared premium.
Cleared Sector Premium
Secret clearance premium over uncleared equivalent: 15% to 25%. Top Secret clearance premium: 25% to 40%. TS/SCI with polygraph: up to 50% in some non-DC markets.
Domain-AI Intersection Premiums
Quantitative finance AI: $400,000+ at hedge funds for senior specialists. Healthcare AI with regulatory compliance: $150,000 to $230,000. Legal technology AI: $150,000 to $220,000. Industrial and manufacturing AI: $140,000 to $220,000.
Roles Under Pressure (for comparison)
Manual QA engineer: $95,000 to $115,000. Traditional programme manager in tech: $110,000 to $150,000. Legacy systems maintenance specialist: $100,000 to $130,000 (compression from prior premium rates).
All figures are for US-based roles in USD, as of early 2026, and represent base salary or total compensation as noted based on the most recently available data from the sources cited throughout this article.
Final note: This article deliberately does not provide links to specific job boards, specific training platforms, or specific companies because those resources change faster than articles can be updated and because the best specific resources for any individual depend on their specific background and target. The frameworks, salary data, and strategic guidance in this article are designed to help you evaluate specific resources you encounter rather than to prescribe specific ones.
