What is a forward deployed engineer, and when do you need one?
The FDE role in plain English: what it is, how it differs from consultants and solutions engineers, and the three ways to get one.
The short version.
- A forward deployed engineer (FDE) is a software engineer who works inside a customer's team and systems and ships production software there.
- The role exists because the hard part of enterprise AI is not the model. It is the data, the workflow, and the legacy systems around it.
- You can hire FDEs in-house, get them from an AI lab, or bring in an independent FDE partner. Each fits a different situation.
The definition
A forward deployed engineer sits with the people who do the work, learns how it actually happens, and builds software against the real data and real systems. The output is not a slide deck or a spec. It is a working system in production.
Palantir made the role famous. In the AI era, OpenAI and Anthropic both run FDE teams to get their models into large customers, and job postings for the title have multiplied.
FDE vs. the roles it gets confused with
| Forward deployed engineer | Solutions engineer | Consultant | Staff augmentation | |
|---|---|---|---|---|
| Main job | Ship a working system in your environment | Help close the sale | Advise and recommend | Add hands you direct |
| Writes production code | Yes | Demos and prototypes | Rarely | Yes |
| Finds the problem | Yes | No | Yes | No |
| Owns the outcome | Yes, through production | No | No | No, you do |
Why FDE matters for AI
- Models are easy to buy. Every company can rent Claude, GPT, or Gemini. Few can make one run a real workflow.
- The gap is integration. Undocumented processes, systems with no API, and messy data block most AI projects, and only someone inside the business sees them.
- Production needs a harness. Evals, guardrails, human approvals, and monitoring are built against live systems, not in a sandbox.
Three ways to get FDEs
- Hire in-house. Best if AI is your product and you can wait months to recruit. You carry the hiring risk and the learning curve.
- An AI lab's FDE team. Deep on one model. The model decision is made before they arrive, and the programs are mostly sized for very large enterprises.
- An independent FDE partner. Senior engineers who embed, stay model-agnostic, and can operate the system after launch. Usually the fastest route for mid-market companies. This is what we do.
Signs you need one
- Your AI pilot works in a demo but has not reached production.
- The workflow you want to automate crosses three or more systems.
- Key software has no usable API, so someone clicks through screens all day.
- Your data cannot leave your own cloud.
- You have no in-house AI team, and no time to build one.
What a good engagement looks like
Embed in week one. Find the costliest workflow in a 4 to 8 week sprint. Engineer the agent and its harness against live systems in 3 to 6 months. Operate it afterwards, or hand it to your team. See real examples in AI in Practice.