Quick test: name the highest-paid software engineering role at OpenAI. Most engineers will guess "research scientist" or "platform engineer." Both wrong. It's Forward Deployed Engineer — a title most people outside AI labs have never heard of, paying $400K to $700K, and one of the fastest-growing tracks in the industry.
The role was invented at Palantir 20 years ago, has been rediscovered by AI labs in the past 18 months, and is rapidly becoming the way every serious AI company sells to enterprise. Here's what FDEs actually do, why they make what they make, and how to become one.
The 30-second answer
A Forward Deployed Engineer is a software engineer who works at the boundary between their company's platform and a customer's environment, building integrations and applications that solve the customer's actual problem. They have full engineering capability — they write code, design systems, and ship features — but their work happens in the customer's domain rather than on a centralized platform team.
Think: engineer with the technical depth of a senior IC, the customer awareness of a solutions architect, and the autonomy of a startup founder.
Where the role comes from
Palantir built the model. The pitch: instead of selling software and leaving customers to integrate it, Palantir sent engineers — actual coders, not consultants — directly into the customer's environment. The FDE figured out the real problem, built a working solution, and either left it running or handed it to the customer's team.
It worked. Palantir's deals became larger and stickier than competitors because customers got something that actually solved their problem, built by people who understood it. Twenty years later, that model is the playbook for selling complex software into enterprises.
Why AI labs adopted the model
By mid-2024, every major AI lab realized the same thing: their best customers had no idea how to use the platform. The technical hand-off wasn't working. Documentation didn't bridge the gap; sales engineers couldn't write production code; consultants didn't have domain depth.
So labs hired FDEs. Anthropic, OpenAI, Sierra, Cohere, Glean, Adept — all of them now have meaningful FDE teams. The job: parachute into a customer's environment, understand what they're trying to solve, and build the system that solves it using the lab's platform as foundation.
What an FDE actually does
A typical engagement runs 6–12 weeks:
- Discovery — meet with the customer's stakeholders, understand the real problem (which is almost never what's in the SOW), identify the data and systems involved.
- Design — sketch the architecture, identify what the platform can do and what needs custom engineering, get alignment.
- Build — write code. Real code. Often in the customer's repo or environment. Often on tight deadlines. With real production constraints.
- Deploy — get it into the customer's production with their team. Handle the gnarly bits: auth, networking, compliance.
- Hand off — train the customer's team to maintain it. Document the system. Move on to the next engagement.
The pace is brutal. The variety is the point. In a year, an FDE might ship in healthcare, finance, defense, and retail — each with different data, different stakeholders, different problems.
Why they make $400K–$700K
FDEs are paid like senior engineers because they are senior engineers, with three additional skill premiums:
- Customer-facing risk tolerance — they sit in front of paying customers and ship under pressure. Most engineers can't.
- Cross-domain fluency — they pick up a new domain in two weeks and ship a system in eight. Most engineers can't.
- Revenue impact — a successful FDE engagement is the difference between a $500K pilot and a $5M expansion. The math on their comp works because each FDE is responsible for millions in ARR.
The role is also rare. Most engineers don't want to travel, don't want customer exposure, don't want the variance. The supply is small; the demand is enormous. Hence the comp.
Companies hiring FDEs in 2026
- Anthropic — large applied / FDE team, focused on enterprise Claude deployments
- OpenAI — "applied AI engineer" and "forward deployed engineer" roles across regions
- Sierra — Bret Taylor's voice-AI company, FDE-heavy go-to-market
- Glean — enterprise AI search, large FDE org
- Cohere — applied teams supporting enterprise deployments
- Adept, Imbue, etc — smaller but growing
- Palantir — the original, still hiring
The roles aren't always titled "Forward Deployed Engineer." Watch for "Applied AI Engineer," "Deployment Engineer," "Solutions Engineer (with code)," and similar — the work is often the same.
What you need to be good at FDE work
Strong engineering fundamentals
You'll write production code under pressure in unfamiliar codebases. Python and TypeScript are the most common; comfortable with REST, queues, databases, cloud infra.
Speed of comprehension
You'll be asked to understand a new domain in days, not months. Read code fast, ask good questions, build a working mental model quickly.
Customer empathy
Engineers without this fail at FDE work. You need to like sitting with customers, listening to their actual problems, and translating those into engineering. Some engineers love it. Many don't.
Comfort with ambiguity
Requirements will change. Stakeholders will disagree. The "real" problem is rarely the one in the brief. You need to be able to operate in this environment without freezing.
Independent execution
FDEs operate semi-autonomously. You'll often be the only engineer on a customer site. Self-directed people thrive; people who need clear instructions struggle.
FDE vs traditional consulting
FDE is not consulting. Consultants advise; FDEs build. Consultants write decks; FDEs commit code. Consultants leave behind reports; FDEs leave behind running systems. The cultures are different, the pay scales are different, and the skill sets are different.
Many consultants try to become FDEs. The ones who succeed are former engineers who went consulting and now want to write code again. The ones who struggle are non-engineers trying to add coding to their toolkit on the job.
How to become an FDE in 2026
- Be a strong software engineer first. The role assumes you can ship. Get there.
- Build customer-facing skills. Even at your current job, volunteer for stakeholder-facing work. Run a demo. Lead a meeting with a non-engineer.
- Pick up AI engineering basics. Most FDE roles now require comfort with LLM platforms — see our AI engineering guide.
- Ship a project end-to-end with someone non-technical. Find a friend in a non-engineering domain, build them something they actually use. That story will sell you in interviews.
- Apply broadly. Many openings aren't titled "FDE." Look for "applied," "deployment," "solutions engineering."
Frequently asked questions
Do FDEs travel a lot?
Some do. At Palantir, heavily. At AI labs, varies — some are remote with periodic customer visits; some are on-site embedded. Ask in interviews.
What's the career trajectory after FDE?
Three common paths: senior FDE → engineering manager of an FDE team; FDE → platform engineering at the same company (now informed by customer reality); FDE → founder (the role builds founder-shaped skills).
Is FDE work going to be replaced by AI?
It's the opposite. As AI capability grows, the bottleneck for deployment is people who can adapt the capability to specific customer environments. FDE demand is growing because of AI, not despite it.
How is FDE different from a regular AI engineer?
AI engineers tend to work on platforms — internal or external — that many users consume. FDEs work on specific customer deployments with intense focus on one customer's needs. See our comparison for the deeper breakdown.
What's the interview like?
Standard senior software engineering interview (coding, system design) plus a customer-facing component — case studies, hypothetical client conversations, sometimes a presentation. The technical bar is the same as a senior software engineer at the company; the customer-facing bar is the differentiator.
Bottom line
Forward Deployed Engineering is a real, demanding, well-paid role that has been quietly absorbing the best engineers in AI. If you're a strong builder who likes customer contact and wants to work on real problems in domains you'd never otherwise see, this is one of the best jobs in technology.
If you want to build the AI engineering fundamentals that make you hire-able for an FDE role, the JoinAI MasterClass ships three production agents over 8 weeks — exactly the kind of project portfolio that opens FDE conversations.
