Enterprises have spent two years piloting generative AI, and most pilots never reached production. The market's answer is a new kind of engineer: one who leaves the vendor's office, sits with the client's team, and ships. That is the Forward Deployed Engineer, and in 2026 everyone from AWS to Deloitte is racing to hire them.
What a Forward Deployed Engineer is, and isn't
An FDE is a software engineer who embeds with a client and builds production systems inside the client's own environment. The role was popularized by Palantir and is now central to how AI labs, cloud providers and consultancies deliver enterprise AI. What sets an FDE apart is owning shipped code in the client's world, and staying until it is adopted.
| Role | Main output | Writes production code? | Stays until adopted? |
|---|---|---|---|
| Forward Deployed Engineer | A working system in the client’s environment | Yes, 70–90% of the time | Yes |
| Solutions or sales engineer | Demos, proofs of concept, the technical win | Sometimes, 30–40% | No, hands off after the sale |
| Management or IT consultant | Strategy, requirements, process design | Rarely | Sometimes |
| Product software engineer | Features for their own company’s product | Yes | No client embedding |
Coding shares from Bloomberry's analysis of about 1,000 FDE postings.
A market that grew more than tenfold in a year
FDE postings grew about 1,165% from 2024 to 2025, and October 2025 was the busiest month on record. By April 2026, one tracker counted more than 1,100 active postings at over 600 companies. The biggest hirers are Google, Deloitte, AWS, Salesforce, EY, OpenAI and Databricks, and consulting firms together post more FDE roles than any product company except Google.
The race reached the top in 2026. AWS committed $1 billion to an FDE unit that starts with hundreds of engineers on fixed, outcome-based pricing. OpenAI runs a separate deployment company built on FDEs, Anthropic has a similar unit, and Google is spending $750 million on a program that embeds its own FDEs with customers.
- Most roles are mid-level. 60% of postings ask for 3 to 5 years of experience. Only 12% are entry level.
- Most are builders. 60% are hands-on builder roles, 30% are sales engineers with extra coding, and 10% build internal tools.
- Travel is part of the job. 68% of postings require it, typically 25% to 50% of the time.
- Regulated industries lead. Where a posting names its customers, financial services, government, healthcare, insurance and energy top the list.
What FDEs actually deliver
FDEs are measured on working software, not slides. Across postings from Anthropic, Deloitte, Palantir and Salesforce, a typical engagement produces:
- Discovery and scoping in working sessions with business owners.
- A prototype on real data, often within days.
- A production system: agents, tools, retrieval and integrations into ERP, CRM and data platforms.
- Guardrails: evaluation, monitoring, human-in-the-loop controls and security review.
- Handover: documentation, runbooks and an exit plan so the client can run it.
The skills the market asks for
Python and production work on LLM agents form the core, followed by cloud and data engineering. These are the share of FDE postings that name each skill:
- Python 66%
- AI agents 35%
- TypeScript 35%
- AWS 32%
- LLM experience 31%
- GCP 22%
- Azure 18%
- Kubernetes 14%
- RAG 12%
Source: Bloomberry, about 1,000 postings. One posting can name several skills.
Senior postings go further: tool use and multi-agent orchestration, MCP servers, evaluation frameworks and LLMOps, end-to-end RAG, data pipelines, enterprise integration, and security and compliance. And 47% of postings explicitly require customer-facing ability. That is the part no coding test measures.
Customer discovery
Turning a vague business problem into a scoped AI use case in a working session.
Translating for every audience
Explaining technical tradeoffs to executives and business users, not only engineers.
High agency in ambiguity
Moving without a spec inside large, complex organizations.
Ownership to adoption
Staying until the system is live and people actually use it.
Domain fluency
Postings increasingly prefer experience in a specific industry or function.
Coaching
Bringing the client’s own engineers along so they can run what was built.
What FDEs cost
The median US base salary in FDE postings is about $174K to $182K. Posted ranges run from $135K–$200K at Palantir and $134.5K–$265K at Deloitte to $200K–$300K at Anthropic. Total pay climbs steeply with equity:
| Employer type | Mid-level total pay | Senior total pay | Equity share |
|---|---|---|---|
| Frontier AI labs | $385K–$510K | $560K–$785K | 60–70% |
| Applied-AI startups (Series B+) | $250K–$340K | $340K–$470K | 45–60% |
| Palantir-style FDSE | ~$215K | $280K–$340K | 40–55% |
| Fortune 500 enterprise AI | $190K–$240K | $240K–$310K | 15–25% |
Estimated US total pay per year, from Perspective AI's 2026 report (self-reported and crowdsourced data; directional). Palantir's median FDE total pay on Levels.fyi is about $233K.
The gap: generalist FDEs in a specialist's world
Read enough postings and a pattern appears. The market hires FDEs for their engineering and hopes they pick up the business on the job. 45% of today's FDEs came from software engineering and only 15% from data science. Just 21% of postings even name the industry they serve.
But an agent that reschedules supply, explains a missed delivery or drafts an S&OP plan is only as good as its builder's grasp of the process. A generalist can learn your tools in weeks. Learning how your planners, buyers and controllers actually decide takes much longer, and that is where most AI projects stall.
The Brainlikes twist
Specialized Forward Deployed Engineers
Functional expert who has worked inside supply chain, operations, sales and finance processes, so discovery starts at the right question.
AI engineer who builds agents, tools and orchestration that hold up in production, in your private cloud.
Data scientist who grounds every agent in your documents and your ERP, CRM and SQL data, with measurable results.
Backed by 20+ years of supply chain and data science work and 80+ delivered projects.
Build or buy: the real cost of hiring your own FDE
Hiring your own FDE is a sound long-term goal. But the cost is not just the salary. You pay through every month the role is open, through the ramp-up, and through the first build, and production value arrives only at the end.
- Time to hire: a senior software engineer takes about 47 days on average from opening the role to an accepted offer. Add a typical notice period and the new hire starts about 2 to 3 months later.
- Recruiting fee: agencies typically charge 15% to 25% of first-year salary, up to 30% for urgent hires.
- Ramp-up: new developers take about 3 months to reach their 10th pull request in large companies. An FDE must also learn your systems, data and processes, so we assume 3 to 6 months before they are fully productive.
- Hidden risks: a mis-hire, or losing a scarce FDE to a better offer, resets the clock.
Illustrative time to first production agent (months)
Hire and ramp your own FDE
About 8 to 12 months before the first agent is in production.
Deploy Brainlikes specialized FDEs
A 2–3 week MVE on your data, then a 4–6 week MVA with one agent in production.
Estimates based on the benchmarks above and our MVE and MVA timelines. Your timeline depends on scope, data access and approvals.
Calculator
What does your own FDE cost before it pays off?
Change the inputs to match your market. Results are estimates, not quotes.
Fully loaded cost per year
–
Spent before the first agent is live
–
Months from today to first agent
–
How we calculate: the recruiting fee plus the loaded monthly cost for the ramp-up and first-build months. The overhead default is our assumption.
The shortcut: trained FDEs now, your own team next
With Brainlikes you skip the hiring queue and the ramp-up. Our specialized FDEs arrive trained, with playbooks, reference architectures and evaluation practices already proven, and they start on your real data in your private cloud. You own everything we build.
Building your own FDE capability is still the right long-term goal, and we help with that too. Most clients combine the two:
- Deploy our FDEs to ship the first agent in weeks.
- Assess your team against the skills the market demands, and map the gaps.
- Upskill your engineers by pairing them with ours on real work, until they run the next project themselves.
You get production value now, and an FDE team of your own without paying for every month of their learning curve.
Sources and method
We used public job postings and published analyses, opened on November 17, 2026. Pay and hiring figures are US-only. Crowdsourced pay data is directional, and figures marked as estimates or assumptions are ours.
- Bloomberry: What I learned analyzing 1K forward deployed engineer jobs (updated Jan 2026)
- FDE Pulse: FDE hiring trends 2026 (Apr 2026)
- Perspective AI: 2026 FDE compensation report (May 2026)
- Levels.fyi: Palantir Forward Deployed Engineer salaries
- Constellation Research: AWS launches $1B forward deployed engineering unit (Jun 30, 2026)
- Anthropic: Forward Deployed Engineer job posting
- Deloitte: Forward Deployed Engineer, Frontier GenAI job posting
- Palantir: Forward Deployed Software Engineer job posting
- Recruiting from Scratch: How long it takes to hire a senior software engineer in 2026
- DX: AI cuts developer onboarding time in half (Sep 2025)
- Dover: Tech recruiter fees cost guide (Nov 2026)