fde
What Is a Forward Deployed Engineer? (2026 Guide)
What a forward deployed engineer does, where the role came from at Palantir, how titles differ across AI companies, and what a working week looks like.

Prepare for the Forward Deployed Engineer loop
A forward deployed engineer (FDE) is a software engineer who works embedded with one customer, writing production code against that customer's data, systems and goals until a product is live and in use. Done means the customer's people changed how they work, and a merged pull request alone does not count.
The role description and job-post details here come from our web research into FDE hiring at eight companies, drawn from 23 public sources including candidate write-ups, interview guides, a Blind thread and Sierra's own engineering blog.
That definition is short. The job it describes is wide, so the rest of this page works through where the title came from, what the work looks like across companies that hire for it, and how to tell from a job post which version you are looking at.
One customer, many capabilities
The title comes from Palantir. Ophy's summary of Palantir's own blog describes the split this way: product developers build "one capability, many customers", while Deltas, Palantir's name for forward deployed software engineers, cover "one customer, many capabilities" (Ophy). Palantir has a second customer-facing role, the Deployment Strategist or Echo, described in the same summary as an analyst and generalist position with 25 to 75 percent travel and Python, R or SQL work.
That phrase, one customer and many capabilities, is the most useful sentence about the role. A product engineer goes deep on one surface and ships it to everyone. An FDE goes wide for one account: some data engineering on Monday, an API integration on Tuesday, a conversation with an operations lead on Wednesday about why the dashboard numbers look wrong to them.
What the job looks like in 2026 postings
The AI companies took the Palantir template and attached it to their own products. The specifics vary enough that reading the posting closely matters more than the title.
Salesforce's lead FDE role for Agentforce, per its posting on Built In, embeds the engineer with one enterprise customer to write the code that ships an Agentforce agent (Built In). Sarvam AI in Bengaluru lists a Forward Deployed Engineer role that prefers experience shipping production agents, working with banking or public-sector customers, and deploying in VPC or on-premise environments; it also lists an FDSE role for its model API and a "Strategic Deployment Engineer" role for air-gapped and classified deployments (Khosla Ventures jobs board). Decagon calls a closely related role Deployment Strategist (Paraform).
The titles drift. Prep guides report the same job advertised as Applied AI Engineer, Agent Engineer, Deployment Strategist or plain AI Engineer (Exponent on Sierra, Aced on Anthropic). Search for the work, not the words.
| Signal in the posting | What to expect |
|---|---|
| Travel percentage stated | Customer-site time is expected; the higher the number, the closer to the Palantir model |
| VPC, on-prem, air-gapped | Regulated customers; expect security and deployment questions in the loop |
| "Production agents", RAG, evals | An AI-lab-style role; expect applied-AI design and an evals question |
| SQL, pipelines, notebooks | A data-platform role; expect data wrangling and a KPI-driven case |
| Quota or revenue target | Closer to sales engineering than FDE; see the comparison linked below |
A week, sketched
What does the work feel like? The sketch below is illustrative, written for this page. The customer is fictional and so is every detail.
On Monday the FDE sits with a fictional insurer's claims team and watches three adjusters work a queue. Nobody asked for this. It is how the FDE learns that adjusters re-key the same policy number into two systems, which no requirements document mentioned. Tuesday goes to reading the policy system's API docs and writing a small client that handles its pagination and rate limits. Wednesday morning, the claims director asks for an AI summary of every incoming claim. The FDE asks what the adjusters would do differently with a summary, and the answer turns out to be "triage faster", which points to a priority score and away from a paragraph of prose. By Thursday a first version ranks the morning queue, measured against how the senior adjuster ranked the same claims. Friday is a short demo, a list of what broke, and a note back to the product team in headquarters: three other insurers will want this same connector.
That last step separates FDE work from consulting. The thita track describes the loop as embed, ship, productize and hand off, and the productize step is the one where something built for one customer becomes a capability for the next (the FDE operating model lesson).
What you need to be good at
The skills split three ways. On the technical side, the research behind our track lists production Python with tests, SQL with joins and window functions, data pipelines and schema mapping, REST integration with auth, pagination and idempotency, LLM application work such as RAG, tool use and evals, and deployment basics like cloud IAM and secrets. On the customer side, it lists discovery, scoping, delivering bad news and explaining an AI system's limits to someone who does not care how embeddings work. On the business side, it lists defining a success metric with a baseline and phasing a rollout so the customer sees value in week one.
Skip the urge to master all of it before applying. The interview loops test a narrower slice, which the FDE interview guide maps company by company, and the question bank shows what that slice looks like in practice.
Is it the right job for you?
That depends on two things: whether you like context-switching across a customer's whole stack, and whether you can stand being the person in the room when something fails. If you want deep ownership of one system for years, a product engineering role fits better. If the idea of watching users work, then changing the code that afternoon, sounds like the good part, read FDE vs software engineer next for a side-by-side.
Start with the free FDE lessons
The FDE track on thita.ai opens with two free patterns: what a forward deployed engineer is and the decomposition framework, plus two worked case studies, improving city traffic with sensor data and a logistics client losing money on deliveries. The full track adds AI mock interviews for five FDE rounds: decomposition, customer discovery, applied-AI design, practical coding and behavioral.
Your next step
Prepare for the Forward Deployed Engineer loop
The FDE track walks through the rounds FDE loops use, decomposition, coding and customer scenarios, then lets you rehearse them out loud with an AI interviewer that asks follow-up questions.