A forward deployed engineer (FDE) is a customer-facing software engineer who embeds with a client's team to build software inside that company's real systems, alongside the people who will use it. The model was popularised by Palantir, which used the title as early as 2009, and the term itself comes from military jargon (Wikipedia).
Where the role comes from, and why it is everywhere now
Palantir developed the FDE as a way of working rather than a job family: engineers start from a defined business outcome and build software against the actual problem, inside the customer's environment (IDC).
Interest in the model has surged recently: McKinsey notes a sharp rise in attention in recent months, and Pragmatic Engineer documented a hiring spree among startups and scaleups back in August 2025. Meanwhile, IDC reports that only 15% of organisations say more than three quarters of their AI projects delivered measurable outcomes.
What a forward deployed engineer actually does
Strip the label away and the work looks like this:
- Scope the real problem. The first weeks are spent sitting with the people who do the work, understanding who decides what, which data matters, and where the hours go (McKinsey's description of a typical first phase).
- Work inside the client's systems. Not in a lab, not on a demo dataset: real data, real tools, real constraints.
- Build, demo, iterate. Working software early, shown to the team, adjusted or restarted based on feedback.
- Change how the work gets done. The tool only matters if the team's workflow changes with it, so training and handover are part of the job, not an extra.
The problem with the label
Here is the part most explainers skip. IDC's assessment is blunt: the FDE label "has outrun a single working definition", and some providers practise "FDE washing", relabelling ordinary contracting or staff augmentation as forward-deployed work. The title tells you almost nothing about what you are buying.
What actually varies between providers:
- Who employs the engineer. A platform vendor, a consultancy or systems integrator, a dedicated deployment company, or an independent agency like ours. This shapes where feedback goes and who the engineer ultimately serves.
- Outcome accountability. Does anyone own a measurable result, or is it hours billed against tasks?
- Commercial structure. Fixed scope and price, monthly retainer, hourly, or a sales call before you hear any number at all.
So instead of comparing titles, compare answers. Five questions cut through any provider's pitch: Will your engineer work inside our systems, on our real data? What exactly ships, and by when? Who owns the business outcome? What happens if it does not ship? And when it works, who owns the code, the documentation and the know-how?
How we run it at Lyra
Our version, in concrete terms. We sit with your team and map where the hours go: the manual, repetitive work AI could take over. We rank the candidates by value, because not every task is worth automating. Then we build in the tools you already use, under your security rules, starting with a fixed-scope Sprint on one workflow. The code, documentation and know-how stay with you. Your team learns to run what we build, so you are not dependent on us afterwards.
With GoCoCo, for example, we moved from an overview of AI opportunities to hands-on consulting with the product and tech team, focused on how they develop and release software.
What a forward deployed engineer costs
Few providers publish prices. One October 2026 review by Eupraxic, itself an FDE provider, found only 9 of 34 providers showing any figure - so treat what follows as a market participant's own compilation, not independent research. From the numbers that are public:
- Fixed-scope sprints and pilots. One agency publishes bands of $8,000-$15,000 for a two-week sprint and $25,000-$90,000 for a six-week production build (forwarddeployedengineers.agency).
- Embedded engineers by the month. Roughly $15,000 to $40,000 per month across the firms that publish, per Eupraxic's compilation.
- Hiring your own. In Eupraxic's review of 382 US job postings, median base salaries ranged from $180,000 to $250,208, before equity.
These are other providers' published figures, collected in October 2026; they move, and most quotes still start with a sales call. But they give you a sanity range for any proposal, including ours.
When you don't need a forward deployed engineer
Honest answer: sometimes you don't. If you already employ strong applied AI engineers, use them. If nobody has decided which process matters most, fix that first; any good provider will tell you the same. And if what you want is a strategy document, buy a strategy document. The model earns its keep when you have a concrete, valuable workflow and you want it working in production, with your team able to run it afterwards.
Frequently asked questions
- Is a forward deployed engineer the same as a contractor?
- Not necessarily, and the label alone will not tell you. Some "FDE" offers are relabelled contracting. The difference to look for is outcome ownership: a real forward-deployed engagement commits to a result in your systems, not hours against tasks.
- How long does an engagement take?
- It varies by provider and scope. Published examples range from two-week sprints to multi-month embedded retainers. What matters more than the calendar is a fixed definition of what ships and when.
- Do forward deployed engineers work on site or remote?
- Both. "Forward deployed" means close to your team and your real systems, not necessarily a desk in your office. Regular working sessions with the people who own the process matter more than location.
- What should we have ready before starting?
- One process your team already finds painful, a person on your side who can make decisions, and a clear path for access: which systems and data the engineer may touch, under what approvals and security review. Scoping the work itself is part of the engagement.
- Who owns the code and IP?
- Ask every provider this before signing, and get it in the contract. At Lyra, the code, documentation and know-how stay with you, as stated on our public offer page.
The short version
A forward deployed engineer is how AI stops being a pilot and starts being a tool your team uses every day. If you have a workflow that eats hours and want to see what it looks like automated, book a demo and we will map it together.