What is a forward deployed engineer?

The role went from a Palantir curiosity to the most contested job in software in about eighteen months. Here is what one actually does, what one costs, and how to decide whether to build the function yourself or bring one in.

The short answer

A forward deployed engineer, usually shortened to FDE, is a senior engineer who works inside the customer's team rather than from the vendor's office. They sit with the people doing the work, write production code in the customer's own systems, and stay accountable for whether the thing actually runs in that environment.

The distinguishing feature is not the seniority or the travel. It is that an FDE is measured on the outcome rather than on delivery of a specification, and is expected to work out when the specification itself is wrong.

01 Definition

How the role differs from the ones next to it.

01 vs consultant

A consultant recommends. An FDE ships.

A consultancy produces an assessment, a roadmap and a set of recommendations, then hands the build to somebody else, often a systems integrator or the client's own team. The engagement is complete when the document is accepted.

A forward deployed engineer writes the production code themselves. The engagement is complete when the system is running and handling real volume. That single difference changes what the person has to be good at: less framework, more debugging somebody else's undocumented database at four in the afternoon.

02 vs solutions engineer

A solutions engineer stops at the demo

Solutions engineers sit in the sales cycle. They scope, they demonstrate, they answer technical objections, and their work largely ends when the contract is signed.

A forward deployed engineer starts roughly where a solutions engineer stops. The demo is the beginning of the job rather than the end of it, and the hard part, the messy data and the exceptions nobody documented, is all still ahead.

03 vs contractor

A contractor builds the spec. An FDE questions it.

A contractor is given a specification and judged on how faithfully they build it. That is a perfectly good arrangement when the requirements are known in advance.

The requirements are rarely known in advance for this kind of work. An FDE is expected to sit with the people doing the job, notice that the documented process and the real process are different, and say so. Being right about what to build matters more than being fast at building it.

04 The profile

Why the profile is so rare

The role sits at an awkward intersection: staff-level engineering ability, the social skill to sit with a claims team for a week without irritating anybody, and comfort working with no specification and incomplete information.

Most strong engineers have the first and not the second. Most client-facing people have the second and not the first. That intersection is small, and it is why the hiring market for it has become what it has.

02 The market

Why hiring one has become so hard.

5,230% Growth in FDE job postings between January 2025 and April 2026, roughly 729 percent year on year
~17,000 Forward deployed engineers in the United States in total, a large share already at Palantir
7 in 10 Companies that said they planned to hire an FDE by the middle of last year

The demand did not appear gradually. Palantir ran the model quietly for years, then generative AI made the gap between a working demonstration and a working system impossible to ignore, and every company selling AI discovered it needed people who could close that gap inside a customer's environment.

OpenAI launched a deployment subsidiary built around forward deployed engineers in May 2026, capitalised at around four billion dollars and seeded by acquiring a consultancy for its 150 engineers. Deloitte and Accenture both stood up dedicated practices. The largest services firms have said they need to increase FDE headcount roughly tenfold, building teams of twenty to a hundred.

All of that demand is competing for the same seventeen thousand people. If you are a four hundred person company trying to hire your first one, that is the market you are bidding into.

03 The decision

Build the function, or bring one in.

Hire when
  • You need a permanent team of several, not one person
  • You have a technical leader who has worked alongside this profile before and can interview for it
  • The work will not run out once the first few workflows are done
  • You can wait three to six months and survive losing the offer
Bring one in when
  • You have one to three workflows to fix rather than a permanent function
  • Nobody internally has hired for this profile and could not assess a candidate
  • There is a deadline that will not survive a six month search
  • You want to prove the motion works before committing to headcount

These are not opposed. A common and sensible sequence is to bring a team in for the first two or three workflows, watch what the work actually involves, and use that to write a job description you could not have written beforehand. By then you know what you are interviewing for.

04 Cost

What one actually costs.

ArrangementWhat it includesTypical
Contract FDEOne person, hourly, you manage them$60 to $250 / hr
Full time hireSalary before recruiter fee, management and hiring risk$200,000+ / yr
A team you bring inEngineers, delivery and usually ongoing operation$95 to $150 / hr

The full time figure understates the real number. Add a recruiter fee of roughly twenty to twenty five percent of first year salary, the cost of whoever manages them, and the fact that a wrong hire in a profile you cannot assess is expensive to discover and slow to correct.

ForwardOps publishes fixed prices rather than quoting hourly: $15,000 for a two week scoping engagement, from $75,000 to build and ship a workflow, from $9,000 a month to operate it afterwards, and from $35,000 a month for a standing function of two to four. The pricing page has the rest.

05 Questions

Common questions.

Who invented the model?

Palantir. It used forward deployed engineers as the mechanism for getting its software working inside government and enterprise environments, at a time when the prevailing wisdom was that software companies should not do services. The bet was that the deployment work was the product. OpenAI, Deloitte and Accenture have all since built practices on the same idea.

Is an FDE the same as staff augmentation?

No, and the difference matters commercially. Staff augmentation supplies people to work through your existing backlog under your direction. A forward deployed engineer takes a problem, decides what should be built, and is accountable for the result. If a firm offers you FDEs but describes ticket throughput, they are selling staff augmentation with better branding.

Can an FDE work remotely?

Partly. The learning phase benefits enormously from being in the room with the people doing the work, because most of what matters is what they do without thinking about it. The build phase is ordinary engineering and travels fine. We tend to be on site for the first week and remote after that.

What should we ask a firm offering FDEs?

Three things. Who writes the production code, the person in the room or somebody offshore you will never meet. What happens after go-live, since most firms leave at exactly the point the system starts to drift. And whether they will tell you not to build something, because a firm that has never talked a client out of a project is not assessing anything.