What is a Forward Deployed Engineer? Role, lifecycle and how FDEs are measured

A forward deployed engineer (FDE) is a software engineer who works inside a customer’s organisation to make a product run on the customer’s real data and systems. According to First Round Review, the FDE embeds with a customer to build the last mile of the product in production, and is still an engineer who writes and debugs production code. One question separates an FDE from a demo engineer. It appears below.

What a forward deployed engineer does day to day

The work is delivery, not advice. An FDE joins the customer’s channels, reads the customer’s data, builds against the customer’s systems and stays accountable for whether the result is used. Palantir, the company that created the role, describes its FDEs as people whose mandate is technical outcomes. According to The Pragmatic Engineer, Palantir places them in a team that supports one customer and measures their success by impact on that customer’s goal.

A current example from an AI lab shows the shape. Anthropic’s FDE job listing says the engineer embeds directly with strategic customers and delivers technical artefacts such as MCP servers, sub-agents and agent skills for production workflows. The same listing mentions possible travel to customer sites, estimated at 25%.

How is that different from a sales engineer or solutions consultant? The code. First Round’s point is that the FDE remains an engineer who writes and debugs production code, where a sales engineer typically demonstrates and a consultant typically advises.

Where the forward deployed engineer role came from

The role began at Palantir. The Pragmatic Engineer reports that the Forward Deployed Software Engineer role was created there in the early 2010s and named “Delta”. Other accounts date it earlier: Netguru places it in the mid-2000s. The dates differ, but the reason is consistent. Netguru says early government and defence clients had data environments too sensitive and idiosyncratic for remote delivery, so a spec handed over by a solutions architect left a ticket queue rather than working software.

The scale was unusual. The Pragmatic Engineer says until around 2016 Palantir had more FDEs than ordinary software engineers. It also notes that after the Foundry platform launched that year, more FDEs moved back into product engineering, bringing field experience to the core product. That loop, from customer site to product, is a defining feature of the role and returns in the measures section.

Why demand for forward deployed engineers has grown

AI changed the economics. The New Stack argues that AI pilots usually fail after the demo, when models meet messy data, undocumented workflows and production systems. Someone has to close that gap, and vendors have started staffing for it. LeadDev reports that FDE job postings rose more than 729% in April 2026 compared with the prior year, according to Indeed data shared with Business Insider.

The money is following. Forbes reports that OpenAI announced alliances pairing its forward deployed engineering team with McKinsey, Boston Consulting Group, Accenture and Capgemini, and that Anthropic announced a joint venture reportedly valued at more than $1.5 billion. Meanwhile AI Chat Daily says AWS launched an FDE organisation and committed $1 billion in internal resources to staff it.

A warning about the title

Treat the growth with some care. LeadDev notes that the title is spreading faster than a clear definition. First Round quotes a Palantir alumna who argues that configuring a product for a customer is not forward deployed engineering without room for discovery and invention. Forbes adds that the economics of the model are unproven outside Palantir. A job ad that says FDE may describe implementation support.

The one question that separates an FDE from a demo engineer

A demo engineer is judged on the demo. The system does what was asked, on tidy data, in front of the people who asked. An FDE is judged after the demo, when the system meets real permissions, odd data and users with their own habits. The New Stack puts it neatly: a model can generate ten plausible answers, and the FDE decides which one the customer needs and adjusts as the customer’s reaction changes the brief.

That judgement shows up in what the system will not do. A system that accepts every request has not met the customer’s real rules. The question is the Refusal Test:

What did your system refuse to build, and what does the refusal look like?

A demo engineer struggles to answer, because nothing in a demo should ever refuse. An FDE answers with specifics: the action blocked for lack of authority, the record rejected as unreliable, the request sent to a person. The answer also shows whether the customer’s rules were understood or merely assumed.

The deployment lifecycle an FDE works through

Good FDE work follows a sequence, and skipping a stage is where deployments stall. AIU teaches it in seven stages:

AIU teaches the deployment lifecycle in seven stages: post-sale handoff, discovery, scoping, build, deploy, iterate and hand-back.

The first stage matters because it fixes where the engineer’s responsibility begins. Netguru describes FDEs as post-sale engineers who close the gap between a signed contract and working code in production. Work therefore starts with what the sales process promised, which is rarely what the customer’s team believes was promised.

Discovery is engineering work, not a meeting series. Anthropic’s FDE listing asks for strong communication skills to conduct discovery with customers alongside Python and production shipping experience.

Hand-back is the stage most teams underplay. When FIS announced an anti-money-laundering agent co-built with embedded Anthropic engineers, The New Stack reported the press release language that the engineers would:

“transfer knowledge so FIS can build and scale additional agents independently over time.”

AWS frames its own FDE work the same way. AI Chat Daily reports that AWS engineers build purpose-built agents inside the customer’s AWS environment and hand off the workflows when the engagement ends. An engagement that leaves the customer dependent has not finished.

How forward deployed engineers are measured

Hours billed and tickets closed say little here. AIU teaches three measures:

AIU teaches that Forward Deployed Engineers are measured on three things: deployment velocity, customer success and product feedback quality.

The wider market points the same way. The New Stack says success is measured by how well the system is used and performs, not by time spent. Palantir’s own framing, quoted above, ties success to the customer’s goal.

Product feedback quality is the measure outsiders overlook. Anthropic’s listing expects the engineer to identify and codify repeatable deployment patterns and contribute insights back to the Product and Engineering teams. That is the Foundry loop again: what an FDE learns in one customer’s building should make the next deployment cheaper. An engineer who ships well but sends nothing back is leaving value behind.

Where the Certified Forward Deployed Engineer credential fits

The Certified Forward Deployed Engineer (CFDE)™ exists because the title has spread faster than a clear definition, and employers and candidates lack a common reference for what the role involves. AIU describes CFDE as built as the first certification standard for the Forward Deployed Engineer role. The course teaches and the examination certifies, and the two are separate.

If you are hiring, use the lifecycle and measures above as interview material. If you are training, the same questions apply: can you scope, build, deploy and hand back a system a customer’s team will actually run?

To see how AIU teaches this and what the programme covers, use the link below.

Explore the CFDE Foundation programme →

Frequently asked questions