What is a Forward Deployed engineer, and how do you hire one?
In May, OpenAI launched a company whose whole product is engineers who go and work inside other companies. The OpenAI Deployment Company started with more than $4 billion behind it and a deal to buy Tomoro, a UK consultancy, for its roughly 150 forward deployed engineers. Days earlier, Anthropic's backers had put $1.5 billion into a venture built on the same idea. In July, Amazon Web Services set up a forward deployed engineering team of its own.
So if you're a founder or CTO trying to hire one forward deployed engineer this quarter, those are the companies bidding against you. Paraform, a US recruiting firm that places these engineers, tells clients to budget 8 to 12 weeks for the search. Plenty of teams don't have 12 weeks, because the enterprise deal that needs the engineer will have closed or died by then.
CareerBuddy keeps a pre-vetted network of US-based forward deployed engineers ready for on-site and hybrid deployment. If you need one, book a call, and we'll send a shortlist matched to your (stack, location, and travel) requirements within 4 to 7 days. If you're still deciding whether you need one at all, everything below is what we tell clients on that first call.
What does a forward deployed engineer do?
A forward deployed engineer (FDE) is a software engineer who works inside a customer's environment and writes production code to make a product work on that customer's real data, systems, and processes. They sit in the customer's Slack, standups, and codebase, and sometimes in their office.
Palantir invented the title. It sent engineers to sit with government and enterprise clients, and until 2016 it employed more FDEs than regular software engineers, according to The Pragmatic Engineer.
In AI, the job usually starts after a demo goes well. The base model works, but turning a prototype into a production-ready LLM deployment requires real engineering. Someone has to build custom agent pipelines, connect them to a 15-year-old core banking system, navigate an undocumented authentication layer, and prove to a compliance team that every API call is secure. That person is the FDE.
Here's how the role compares with the two jobs it gets confused with:
A forward deployed engineer works in one customer's live environment, writes production code in that customer's stack, deals with the customer daily, and is judged on whether the customer ends up running in production.
A software engineer works on your core product, writes production code in your own codebase, deals with customers occasionally, and is judged on the features they ship.
A solutions engineer works on demos and architecture before the sale, rarely writes production code, deals with customers daily, and is judged on deals won.
The title is also splintering, which matters when you go looking for candidates. Perspective AI went through about 1,000 FDE job posts and found the same job listed as Forward Deployed AI Engineer, Applied AI Engineer and Deployment Solutions Engineer. By its estimate, roughly a third of the market sits under those other titles.
Why demand for forward deployed engineers jumped in 2026
FDE job postings on Indeed in April 2026 were about 5,230% above January 2025 levels, roughly 729% up year on year (Indeed data, reported by Business Insider).
Job postings for forward deployed engineers from January to August 2026 were up more than 1,000% on the same months of 2025, against 13% for all tech postings (Lightcast, via Fortune).
FDE postings in India in June 2026 were nearly 19 times the year before (Indeed India).
The OpenAI Deployment Company launched in May 2026 with more than $4 billion, bought Tomoro, then bought Northslope in July (OpenAI, Forbes).
An Anthropic-backed enterprise venture raised $1.5 billion and bought Fractional AI in May and Casper Studios in August (Decrypt, Forbes).
The money is chasing a plain problem: AI pilots stall at the point where they have to touch real systems. S&P Global's 2025 research found that the share of companies abandoning most of their AI initiatives before production rose from 17% to 42%.
Here's what that looks like up close. Gritmind, a software firm in Kaunas, Lithuania, described an invoice-validation project to Forbes in September. The client's staff decided whether an invoice was valid using what they knew about vendor contracts and past transactions, and most of that knowledge wasn't written down anywhere. An AI agent can't follow a rule nobody wrote. Someone has to sit with the accounts team, find the rules, turn them into code, and decide which calls stay with a person. That's FDE work, whatever the job title says.
Why forward deployed engineers are hard to hire
The first problem is the skill mix. You need someone who ships production code and can also hold a room with a sceptical IT director. Paraform is blunt about this in its guidance to clients: most strong backend engineers have no interest in customer-facing work, and most consultants who are good with clients can't ship production systems. The people who can do both are a small group, and the AI labs are calling them first.
The second is price.
Here's what a US hire looks like in 2026:
Median advertised base salary: $173,816 (Bloomberry, analysis of 1,000 FDE job posts)
Median total compensation: $220,000 (Levels.fyi)
OpenAI's San Francisco base band: $185,000 to $300,000, plus equity (OpenAI job posting, via Paraform)
Recruiter fee: about 25% of first-year base, or $45,000 on a $180,000 salary (Paraform)
Time to hire: 8 to 12 weeks (Paraform)
The third is the interview. Most companies run their standard software engineering loop, which tests algorithms and system design in a vacuum. It passes people who'd hate sitting with customers and fails people who'd be excellent at it. Paraform also says the most common reason an FDE search drags is that the hiring team never agreed which kind of FDE it wanted.
How to hire a forward deployed engineer, step by step
Check that you need one. First Round Review has reported that founders who come asking for an FDE often turn out to need a software engineer or a post-sales hire. The clearest sign you do need one: an enterprise proof of concept has stalled on integration, and pulling your core product engineers onto it would stall your roadmap.
