Forward-Deployed Engineer job description (with template)
What the role is actually responsible for, the requirements that predict success, and a JD you can copy — written by someone who interviews against it.
A Forward-Deployed Engineer is a customer-facing engineer who deploys, integrates, and customizes a company's product directly inside client environments — increasingly for AI/LLM products. The job description splits the same way the job does: half hard engineering (integrations, data plumbing, LLM apps, evals), half field work (discovery, scoping, demos, stakeholder management). Below: the responsibilities and requirements that belong in the posting, a copy-paste template, and the alternate titles the role hides under — useful whether you're writing the JD or deciding if a posting is really this job.
Core responsibilities
- Discovery. Sit with customer stakeholders and dig past what they asked for to the problem they actually have.
- Integration engineering. Wire the product into the customer's stack — their systems, auth, and data formats — and handle the messy data that comes with it.
- Deploy and demo. Get working software live in the customer's environment and show it to the people who use it, on their terms, in their language.
- The AI/LLM layer (AI companies). Ground the product in the customer's documents and data (retrieval), measure whether answers are right (evals), and handle the failure cases (guardrails).
- Field-to-product feedback. Translate what breaks and what's missing at the account into signal the product team can build against.
Requirements that actually predict success
- Solid fundamentals in at least one mainstream language — FDEs debug alone inside customer environments, so this is the floor, not a differentiator.
- Proven end-to-end shipping under real constraints: took something from unclear requirement to running software, fast.
- Comfort with APIs, webhooks, integrations, and data that arrives broken.
- Clear communication with non-technical stakeholders — can explain a technical situation in the customer's terms.
- For AI products: hands-on LLM application work (RAG, evals, guardrails). Building with models, not researching them.
One requirement to leave out: "X years of experience as a Forward-Deployed Engineer." The title is too new — almost nobody has it, and the people who'd clear your bar did the equivalent work under other names: consulting delivery, internal tools with real internal "customers," full-stack ownership. Requiring the title filters out exactly the pool you want (candidates: this is why you should apply anyway).
The template
Copy, then swap the bracketed parts:
Forward-Deployed Engineer — [Company]
[Company] builds [product, one line]. Our customers' environments are where the product succeeds or fails, and that's where you'll work.
What you'll do: Embed with customer accounts from kickoff to live deployment. Run discovery with their stakeholders, then build: integrations against their systems, pipelines over their (messy) data[, and LLM features grounded in their documents — with evals to prove they work]. Demo what you build to the people who use it. Bring what you learn in the field back to our product team.
What we're looking for: An engineer who ships end-to-end under ambiguity. Strong fundamentals in [languages]; comfort with APIs, integrations, and data that arrives broken; the ability to explain technical decisions to non-technical stakeholders[; hands-on experience building LLM applications]. You don't need to have held the FDE title — internal tools, consulting, and full-stack product work all count if you owned the outcome.
How we interview: A time-boxed practical build, a client-conversation exercise, and an engineering-fundamentals conversation. No puzzle algorithms.
The "how we interview" line is deliberate: it tells strong field-shaped candidates your loop tests the job, which is a filter in itself. If you're designing that loop, what to look for when hiring FDEs covers the three signals and the red flags.
The titles this role hides under
The title isn't standardized yet, so the same JD ships under several names. Post under (or search for) any of these:
- Forward-Deployed Engineer / Forward Deployed Software Engineer
- Solutions Engineer / Solutions Architect — when the posting emphasizes building, not demos (the difference)
- Deployment Engineer / Implementation Engineer
- Applied AI Engineer / AI Engineer, customer-facing
- Member of Technical Staff — Deployed / Field Engineering
Whichever direction you're reading this from, the test is the posting body, not the title: "work directly with customers," "ambiguous environments," and "ship quickly" together mean it's this job.
Writing the JD — or answering one?
The free field guide covers the 3 traits FDE loops actually score, from someone who runs them. Useful on both sides of the table.