Forward Deployed Engineer (FDE)
rolesFDE · embedded engineer · forward deployed engineering
A technical profile embedded inside the client's team to take an AI use case from demo to a system running in production. Not a consultant, not an in-house developer: the bridge between a model that already works and an operation that still doesn't use it.
The Forward Deployed Engineer (FDE) is the market's answer to a problem technology no longer causes: the pilot works, the demo convinces, and still nobody uses the system on Monday morning. The FDE sits inside the client's operation, redesigns the real workflow, integrates with the systems that already exist, defines acceptance criteria, and does not leave until the process runs in production on real data. What they actually do: (1) pick the use case by operational value, not by demo appeal; (2) redesign the process before automating it; (3) integrate with whatever CRM, ERP, inbox or proprietary system is there; (4) build the evaluations and the human-in-the-loop that make the system auditable; (5) hand over documentation and transfer the operation to someone inside. When you need one: when technology is not your bottleneck. If your pilot passed the test but never reached production, that gap is exactly theirs. If you still don't know which process to automate, an FDE is expensive and premature. The nuance almost nobody mentions: the enterprise version of this role is a dedicated squad for months, usually tied to one specific platform. A 50-to-500-person company can rarely hire that in-house or afford that model. The realistic alternative isn't giving up the function: it's buying it in fractions, tied to a specific workflow rather than to a platform.
How it differs from
- AI consultant
- The consultant delivers recommendations and a plan; the FDE delivers a running system. The consultant's metric is the report; the FDE's is that the process works without them.
- ML Engineer / AI Engineer
- The AI Engineer builds the technical capability; the FDE lands it in one client's actual operation, with their data, their permissions and their people.
- Customer Success Engineer
- The CSE handles adoption and incidents on an already-deployed product; the FDE builds the deployment that doesn't exist yet.
- AI Process Owner
- The process owner is internal and stays forever; the FDE comes from outside and succeeds by being able to leave.
Examples
- An FDE spends three weeks inside procurement, rebuilds the approval loop and leaves an agent matching invoices against purchase orders in production.
- An FDE kills the use case leadership asked for because the source data doesn't exist, and proposes one that can actually be measured.
FAQ
- How is an FDE different from hiring a consultancy?
- The deliverable. A classic consultancy delivers diagnosis, roadmap and governance; the FDE delivers the process running in production. You can buy both, but read the invoice: if you're paying for work done and you receive a slide deck, you did not buy an FDE.
- Do I need an FDE if my company has 80 people?
- You need the function, not necessarily the headcount. In a mid-sized company the role is usually covered by an external profile assigned to one specific workflow for a few weeks, with written acceptance criteria and handover to someone internal. Hiring full-time only makes sense with a continuous pipeline of processes to automate.
- What should I ask an FDE before signing?
- Three things in writing: which specific process will end up in production, how you'll measure that it works (evaluations, not impressions), and who on your team operates it once they leave. If any of the three has no answer, you're still buying a pilot.
- Is this a new role?
- The name spread through the applied-enterprise-AI ecosystem during 2025-2026, once it was clear the bottleneck was landing, not modelling. The function is older: there has always been someone who moves into the client's operation until the system starts. What's new is that it now has a name, a market and a price.