By FDE Interview Prep, a SourceStrongAI company

In this post11 sections
  1. The one-line difference
  2. What each posting asks for
  3. Side by side, on one screen
  4. Where the titles blur: agent engineers and applied AI engineers
  5. Is one growing faster?
  6. Which pays more?
  7. What each interview tests
  8. How to choose, and how to move between them
  9. Questions people ask
  10. Keep reading
  11. More from the blog

You build with LLMs every day, and a recruiter has just sent you a role. Or you hold an AI engineer offer and an FDE offer, and the two job descriptions read almost the same. The short answer: both build with the same models, but the FDE owns one customer’s outcome inside that customer’s systems, while an AI engineer on a product team builds for their own company’s product and users. For the wider comparison with software engineering, read the FDE vs software engineer guide.

Below: real postings side by side, the titles that blur the line, what each interview probes, and how to choose.

The one-line difference

One customer, many capabilities; or one capability, many customers. Palantir put the split that way: a software engineer’s focus is “one capability, many customers”, and a forward deployed engineer’s is “one customer, many capabilities”. Source 1Dev versus Delta: Demystifying engineering roles at PalantirPublisherPalantir BlogSource typecompany blogSource 2A Day in the Life of a Palantir Forward Deployed Software EngineerPublisherPalantir BlogSource typecompany blog Swap “capability” for “LLM feature” and that is the FDE vs AI engineer difference.

Both jobs start from the same model APIs. What differs is who the user is, where the code runs, and who answers for the result.

  • The FDE works for one customer at a time, inside that customer’s data, permissions and systems, and is judged by whether that customer’s system works in production.
  • The AI engineer on a product team builds LLM features into their own company’s product, in their own codebase, for every user at once.

The FDE half has a clear source. Sierra’s Natalie Meurer said at the AI Engineer World’s Fair, in a talk posted in July 2026, that across the many versions of the FDE role, the one constant is that every forward-deployed engineer is accountable to the customer. Source 3The Dirty Secret of Forward Deployed Engineering (Natalie Meurer, Head of Agent Engineering, Sierra; AI Engineer World's Fair 2026)PublisherAI EngineerSource typerecorded talk or interview

The AI engineer bullet is our summary: the title covers too many jobs for one definition.

The Pragmatic Engineer, in an August 2025 piece, names engineer, , and technical delivery engineer as close to the FDE, and says the difference is that FDEs work with customers and also contribute to the product their company sells. Source 4Forward Deployed Engineers (Gergely Orosz)PublisherThe Pragmatic EngineerSource typenews report Keep that last part in mind: an FDE is not only a consultant with an API key.

What each posting asks for

Titles mislead; duties don’t. Here is what the postings say.

The FDE posting

As of September 2026, OpenAI’s FDE posting in San Francisco says FDEs own discovery, technical scoping, system design, build and production rollout with strategic customers, and asks for people who write and review production-grade code. Source 5Forward Deployed Engineer (FDE) - SFPublisherOpenAI (Ashby)Source typecompany job posting It also requires travel of up to 50%. Source 5Forward Deployed Engineer (FDE) - SFPublisherOpenAI (Ashby)Source typecompany job posting OpenAI’s financial services FDE posting in New York adds on-call readiness for the production AI systems the FDE ships. Source 6Forward Deployed Engineer (FDE), Financial Services- NYCPublisherOpenAI (Ashby)Source typecompany job posting

Anthropic’s FDE posting places the role on the Applied AI team and asks FDEs to work within customer systems to build production applications with Claude models, with estimated potential travel of 25% to customer sites. Source 7Job Application for Forward Deployed Engineer at AnthropicPublisherAnthropic (Greenhouse)Source typecompany job posting Its FDE postings also name the deliverables: servers, sub-agents and agent skills used in production workflows. Source 7Job Application for Forward Deployed Engineer at AnthropicPublisherAnthropic (Greenhouse)Source typecompany job posting

The applied AI engineer posting at a lab

OpenAI’s Applied AI Engineer posting says the role writes and debugs code and works hands-on to build prototypes, evaluation harnesses, reference implementations, integrations and production accelerators. Source 8Applied AI EngineerPublisherOpenAI (Ashby)Source typecompany job posting On OpenAI’s job board, the FDE posting sits in the Forward Deployed Engineering department, while most Applied AI Engineer postings sit in Go To Market, on the Technical Success team. Source 5Forward Deployed Engineer (FDE) - SFPublisherOpenAI (Ashby)Source typecompany job postingSource 9Applied AI ArchitectPublisherOpenAI (Ashby)Source typecompany job posting

Anthropic’s Applied AI Engineer (Enterprise Tech) posting describes a technical advisor to customers building on the Claude API, who works with account executives and Applied AI Architects and develops customized pilots, prototypes and evaluation suites. Source 10Job Application for Applied AI Engineer, Enterprise Tech at AnthropicPublisherAnthropic (Greenhouse)Source typecompany job posting Its Munich “Applied AI Engineer, Enterprise” posting goes further: once a customer decides to build on Claude, the engineer sets up the customer’s systems integration partner to deliver, “while the partner does the build.” Source 11Applied AI Engineer, Enterprise (Munich, Germany)PublisherAnthropic (Greenhouse job board)Source typecompany job posting

The AI engineer on a product team

This one is our framing, drawn from what the job involves rather than from a cited posting. You build retrieval, prompts, evaluations and agent tooling into your own company’s product. Your user reaches your work through that product, your code ships through your company’s release process, and a product manager helps set priorities. A customer’s strange data reaches you through the product’s support process, not through a week on their site.

Read the verbs

Put the three side by side and the verbs tell you the job:

  • FDE postings: own, scope, build, roll out, be on call.
  • Lab applied AI postings: advise, prototype, evaluate, hand over.
  • Product AI engineering: ship, measure, iterate, scale.

All three write code. The question that separates them is who owns the build after the demo.

Side by side, on one screen

Our summary of the two roles the query asks about, drawn from the postings above for the FDE and from our framing for the AI engineer.

The FDEThe AI engineer
User: one customer’s staffUser: your company’s users
Code runs: in the customer’s environmentCode runs: in your own product
Owns: one outcome, discovery to rolloutOwns: a feature or system, across releases
Hard part: messy data, access, peopleHard part: quality at scale, latency, cost
Travel: up to 50% (OpenAI SF) Source 5Forward Deployed Engineer (FDE) - SFPublisherOpenAI (Ashby)Source typecompany job posting, est. 25% (Anthropic) Source 7Job Application for Forward Deployed Engineer at AnthropicPublisherAnthropic (Greenhouse)Source typecompany job postingTravel: rarely the point of the job
Judged on: the customer’s resultJudged on: product metrics

The third role, the lab’s applied AI engineer, sits between them:

  • User: the customer’s engineers, who build on the lab’s API.
  • Work: pilots, prototypes, evaluation suites and technical advice.
  • Who builds for production: often the customer or a partner. Anthropic’s Munich posting says the partner does the build. Source 11Applied AI Engineer, Enterprise (Munich, Germany)PublisherAnthropic (Greenhouse job board)Source typecompany job posting

Where the titles blur: agent engineers and applied AI engineers

If you only search “forward deployed engineer”, you will miss some of the closest matches. And some roles titled “AI engineer” are FDE jobs.

Sierra’s agent engineers. In September 2026, no job title on Sierra’s main board used “Forward Deployed Engineer”; its customer-facing role for building agents is titled Software Engineer, Agent, on the Agent Engineering team. Source 12Software Engineer, AgentPublisherSierra (Ashby)Source typecompany job posting Sierra defines agent engineers as people who work with its customers to design, build and ship agents on its Agent OS platform, and its postings say these agents run in production and are not prototypes. Source 13Meet the AI agent engineer (Natalie Meurer)PublisherSierraSource typecompany blogSource 14Software Engineer, AgentPublisherSierra (Ashby)Source typecompany job posting In a January 2025 Sierra video, product manager Zack Reneau-Wedeen described the agent engineer as a combination of three jobs: product engineer, frontier-AI builder, and someone deeply engaged in customer success. Source 15What is the role of the Agent Engineer at Sierra?PublisherSierra (YouTube)Source typerecorded talk or interview Our read: that is an FDE description under another name. If you apply there, Sierra’s AI-native interview walks through the loop.

Cognition’s Applied AI Engineer. The posting says the role deploys Devin rather than demoing it (“You don’t demo Devin - you deploy it”), and includes leading live workshops and pair-programming sessions for enterprise engineering teams. Source 16Applied AI Engineer @ CognitionPublisherCognition (Ashby job board)Source typecompany job posting Same title as the lab roles above, much closer to deployment.

“AI engineer” inside an FDE title. Databricks posts a role titled “AI Engineer – Forward Deployed Engineering (AI FDE)”. Source 17AI Engineer – Forward Deployed Engineering (AI FDE)PublisherDatabricks (careers site / Greenhouse)Source typecompany job posting It describes that team as customer-facing, delivering professional services engagements that help customers build and productionize AI applications. Source 17AI Engineer – Forward Deployed Engineering (AI FDE)PublisherDatabricks (careers site / Greenhouse)Source typecompany job posting Palantir posts a Forward Deployed AI Engineer who owns Gen AI strategy and implementation with customers, and asks for past experience building solutions with LLMs. Source 18Forward Deployed AI Engineer (New York, NY) - Palantir on LeverPublisherPalantir Technologies (Lever)Source typecompany job posting

Both, in one pod. Salesforce’s Agentforce Orchestration FDE posting places the FDE in a deployment pod alongside a and Specialized AI Engineers. Source 19Forward Deployed Engineer (FDE) — Agentforce OrchestrationPublisherSalesforce (Workday careers site)Source typecompany job posting Here the two titles are teammates, not alternatives.

Our test for any posting

Ignore the title and answer these from the posting itself:

  1. Which department is it in? At OpenAI this settles it: the FDE posting sits in Forward Deployed Engineering, most Applied AI Engineer postings in Go To Market. Source 5Forward Deployed Engineer (FDE) - SFPublisherOpenAI (Ashby)Source typecompany job postingSource 9Applied AI ArchitectPublisherOpenAI (Ashby)Source typecompany job posting At Anthropic it does not: the FDE sits on the Applied AI team Source 7Job Application for Forward Deployed Engineer at AnthropicPublisherAnthropic (Greenhouse)Source typecompany job posting, and so does the Munich Applied AI Engineer. Source 11Applied AI Engineer, Enterprise (Munich, Germany)PublisherAnthropic (Greenhouse job board)Source typecompany job posting When the department is shared, question 2 decides.
  2. Who owns the build after the demo? “Deploy”, “production rollout” and “on call” point to FDE work. “Prototype”, “pilot” and “advise” point to the advisory side.
  3. Where does the code run? In the customer’s environment, in your company’s product, or in a demo repo the customer extends.

The lesson on reading an FDE job posting line by line takes two real postings apart this way, and who hires FDEs, and under which titles lists the other names the role goes by. If the posting looks like pre-sales wearing an FDE title, FDE or a solutions architect role in disguise? covers the lines to look for, and if it reads more like account support, FDE vs technical account manager draws that line.

Is one growing faster?

A LinkedIn study, as reported by Computerworld in May 2026, found FDE positions grew 42-fold between 2023 and 2025, against 13-fold for AI engineer jobs. Source 20Here’s one career emerging from the AI shift: ‘forward-deployed engineers’PublisherComputerworld (Agam Shah)Source typenews reportSource 21Why Forward Deployed Engineers Are Becoming the Delivery Layer for Enterprise AIPublisherTechScoop (Substack)Source typeinterview prep site

Read that carefully:

So growth is a weak reason to pick one. Pick by the work you want to do every day.

Which pays more?

No dataset we have compares the two titles like for like, and posted bands vary by company, level and city. The FDE salary guide sets out what employers post and file.

What each interview tests

We have found no employer that publishes how it weights its rounds, so what follows is our method, not a rubric: prepare from the verbs in the posting, since those are the duties you will be hired to do.

For an FDE loop, the posting’s “discovery, scoping, rollout” means you should expect to be handed a vague customer problem and asked what you would ask before you design anything, then to write working code. Some loops make the customer live: one Blind poster reported, in July 2026, that a Google FDE (L4) architecture round was a role-play with the interviewer as a company’s CTO. Source 22FDE Interview Experience at Google (L4) - BlindPublisherBlindSource typecandidate report on BlindSource 23Is Google FDE interview same as SWE? - BlindPublisherBlindSource typecandidate report on Blind Practice scoping a first slice before you design anything; the lesson on walking skeletons shows how.

For an applied AI or product AI engineer loop, “evaluation harnesses” and “evaluation suites” in the posting mean you should be ready for a question like a customer asks how you know your AI system works. The post how do you know your AI system works? gives a complete answer, with the words to use. Also rehearse when you would fine-tune, retrieve or just prompt.

For both, expect questions about the models themselves, such as which of our APIs have you built with, and what did you find limiting?

One prompt, two answers

Try this prompt: “A support team wants an LLM to draft replies to their tickets.” Here is how the two answers differ.

The FDE answer starts with the customer. Where do the tickets live, and can I get read access this week? Who approves a draft before it goes out? What does a wrong answer cost them: an annoyed user, or a refund? What would their head of support call a win in a month? Then: one queue, one ticket type, drafts shown to agents and never sent automatically, running in their helpdesk by Friday.

The AI engineer answer starts with the product. How does this work for every customer, not one? What eval set tells us drafts are good enough to ship? What latency and cost per draft can we afford? How do we roll out behind a flag and watch the metrics?

Both answers are good, in the right room. The mistake to avoid is giving the second answer in an FDE loop: designing the platform when the interviewer wanted one customer’s first working slice. The fix is to name the customer, the first user and the first slice before you say the word “platform”. To practice the FDE answer against a customer who pushes back, run the free practice case.

The words for “why FDE and not AI engineering?”

You will be asked some version of this. The question why FDE and not a software engineering role? has a framework and a model answer. For an AI engineer making the move, use its three parts: one piece of evidence, the reason as a choice about the work, and the cost in one clause. An illustrative answer about a fictional team:

“On my current team I build the retrieval layer for our support product. The best week I had was the one I spent with a single enterprise customer whose contracts broke our chunking: retrieval found the right clause for six of their twenty test questions. I rebuilt the chunking with their legal team, got it to seventeen, and we shipped the fix to every customer. I want that week to be the job: owning one customer’s result in their systems, and writing the code that gets it there. I know that means travel and less depth in one codebase, and I want that trade.”

Tell your own version, from your own work. The bridge story lesson shows how to build it.

How to choose, and how to move between them

This part is ours. Choose by who you want your user to be.

The FDE is likely your job if

  • You want to see the person who uses what you built.
  • Messy data, missing access and unclear owners sound like the interesting part.
  • You are fine with the travel the posting states.
  • You would rather own one result end to end than one component across many.

The AI engineer is likely your job if

  • You want to go deep on one product and one codebase over years.
  • Model quality at scale, latency and cost are the problems you enjoy.
  • You want a stable team and a roadmap, not a new customer each quarter.

At least one employer accepts either background. Cognition’s Applied AI Engineer posting asks for 3+ years as a software engineer, technical consultant, deployment strategist, forward deployed engineer, solutions engineer or similar. Source 16Applied AI Engineer @ CognitionPublisherCognition (Ashby job board)Source typecompany job posting Jia Wu, a deployed engineering lead at Cognition, said at the AI Engineer World’s Fair, in a talk posted in July 2026, that business sense can be learned, while being the technical expert in the room is very hard to teach. Source 24How Forward Deployed Engineering is done at Cognition (Jia Wu; AI Engineer World's Fair 2026)PublisherAI EngineerSource typerecorded talk or interview

If you are an AI engineer moving to FDE, your model skills carry over. What you need to show is the customer half: scoping with a real person, saying no to scope, and explaining what the model gets wrong to someone who is not an engineer. If you are an FDE moving to a product AI role, show the depth half: evaluations you built, and systems you kept running after launch.

Still not sure which one you are? Start with Module 0. The FDE role, decoded is free: take the posting you have open, run it through the classification exercise in telling the FDE apart from its neighbors, then build your why-FDE answer with the bridge story.

GlossaryForward deployed engineerA software engineer who builds and ships production systems inside a customer’s problem and environment, accountable to that customer’s outcome.More on Forward deployed engineerGlossaryAgentA system in which a model chooses steps and tool calls to complete a task, within limits the design sets.More on AgentGlossarySolutions architectA customer-facing technical role that designs, advises and often builds proofs of concept; postings place it before the sale, after it, or both.More on Solutions architectGlossarySales engineerA technical role on or beside a sales team that helps win deals with demos and proofs of concept; the U.S. Bureau of Labor Statistics says employers usually pay it salary plus commissions or bonuses.More on Sales engineerGlossaryModel Context ProtocolAn open protocol for exposing tools and data to models through servers, so agents can call a customer’s systems in a standard way.More on Model Context ProtocolGlossaryDeployment strategistA customer-facing role that works out the customer’s questions and scope beside FDEs; at some employers the title means a product-manager or quota-carrying role instead.More on Deployment strategist

Questions people ask

What is the difference between an FDE and an AI engineer?

An FDE owns a customer’s outcome inside the customer’s systems: OpenAI’s FDE posting says FDEs own discovery, technical scoping, system design, build and production rollout with strategic customers. An AI engineer on a product team builds for their own company’s product and users.Source 5Forward Deployed Engineer (FDE) - SFPublisherOpenAI (Ashby)Source typecompany job posting

Is an Applied AI Engineer the same as an FDE?

Not at OpenAI or Anthropic. On OpenAI’s job board, the FDE posting sits in a Forward Deployed Engineering department, while most Applied AI Engineer postings sit in Go To Market. Anthropic’s Applied AI Engineer (Enterprise Tech) posting describes a technical advisor to customers building on the Claude API.Source 5Forward Deployed Engineer (FDE) - SFPublisherOpenAI (Ashby)Source typecompany job postingSource 9Applied AI ArchitectPublisherOpenAI (Ashby)Source typecompany job postingSource 10Job Application for Applied AI Engineer, Enterprise Tech at AnthropicPublisherAnthropic (Greenhouse)Source typecompany job posting

What do news reports say about FDE vs AI engineer job growth?

A LinkedIn study published earlier in 2026, as reported by Computerworld, found FDE positions grew 42-fold between 2023 and 2025, against 13-fold for AI engineer jobs. That is one dataset: do not compare it with growth figures from other sources.Source 20Here’s one career emerging from the AI shift: ‘forward-deployed engineers’PublisherComputerworld (Agam Shah)Source typenews reportSource 21Why Forward Deployed Engineers Are Becoming the Delivery Layer for Enterprise AIPublisherTechScoop (Substack)Source typeinterview prep site

Keep reading

Guides

Lessons

Questions

More in The FDE role