“So what do you think a does?” The weak answer sounds safe: “a hybrid of engineer and consultant who works closely with customers.” Palantir’s blog calls it “a common misconception” that a Delta is a consultant, and that answer never says you ship anything. Source 1Dev versus Delta: Demystifying engineering roles at PalantirPublisherPalantir BlogSource typecompany blog This lesson replaces it with three sentences built from the posting in front of you, one line on how that company measures the work, and a test for whether you would configure a platform or build bespoke systems.

The definition employers write down

  • Palantir’s careers page, as of September 2026, calls its Forward Deployed Software Engineers “Deltas” and gives them a mandate “to achieve technical and operational outcomes for our customers.” Source 2Palantir Careers | Students and Early TalentPublisherPalantir TechnologiesSource typecompany hiring page An named Brian put it more plainly on Palantir’s blog in 2020: a software engineer “who embeds directly with our customers to configure Palantir’s existing software platforms.” Source 3A Day in the Life of a Palantir Forward Deployed Software EngineerPublisherPalantir BlogSource typecompany blog
  • OpenAI’s San Francisco FDE posting, as of September 2026, says the team “partners with customers to turn research breakthroughs into production systems.” Source 4Forward Deployed Engineer (FDE) - SFPublisherOpenAI CareersSource typecompany job posting The same posting lists what the engineer owns: discovery, technical scoping, system design, build and production rollout. Source 4Forward Deployed Engineer (FDE) - SFPublisherOpenAI CareersSource typecompany job posting
  • Anthropic’s FDE postings put the role on its Applied AI team, embedded with the company’s most strategic customers to build production applications with Claude, and name what it delivers: servers, sub-agents and skills “that will be used in production workflows.” Source 5Forward Deployed Engineer (New York City, NY; San Francisco, CA; Seattle, WA)PublisherAnthropic (Greenhouse job board)Source typecompany job posting
  • Google Cloud’s GenAI FDE postings say: “Unlike traditional advisory roles, you will function as an innovator-builder, moving beyond high-level architecture to code, debug, and jointly ship bespoke agentic solutions directly within the customer’s environment.” Source 6Forward Deployed Engineer IV, GenAI, Google Cloud — Google CareersPublisherGoogleSource typecompany job posting

Each puts the engineer on the customer’s problem, building or configuring software that reaches production. From those lines, build a three-sentence answer:

A forward deployed engineer is a software engineer who works inside one customer’s problem and ships production software there, not slides. Your posting puts it as owning everything from discovery to production rollout, which I read as: I am the one who gets paged when the rollout breaks, not the one who hands a design to someone else. The work is judged on whether the customer’s workflow changes, and what the field learns goes back into the product.

That middle sentence comes from OpenAI’s posting. Rewrite it for each company: quote their posting, then say what it commits you to. For an AI lab: “Anthropic’s posting says the MCP servers and agent skills its FDEs deliver ‘will be used in production workflows’, which I read as: I own them after launch, not just the demo.” Source 5Forward Deployed Engineer (New York City, NY; San Francisco, CA; Seattle, WA)PublisherAnthropic (Greenhouse job board)Source typecompany job posting For an enterprise platform company: “Palantir’s careers page gives Deltas a mandate of ‘technical and operational outcomes’ for the customer, which I read as: it counts when their operation runs differently, not when the pipeline is green.” Source 2Palantir Careers | Students and Early TalentPublisherPalantir TechnologiesSource typecompany hiring page Then do the same with the posting you applied to.

One customer, many capabilities

Palantir’s careers page, like its older blog, contrasts the roles in two phrases: a Dev’s focus is “one capability, many customers,” while a Delta’s focus is “one customer, many capabilities.” Source 1Dev versus Delta: Demystifying engineering roles at PalantirPublisherPalantir BlogSource typecompany blogSource 2Palantir Careers | Students and Early TalentPublisherPalantir TechnologiesSource typecompany hiring pageSource 3A Day in the Life of a Palantir Forward Deployed Software EngineerPublisherPalantir BlogSource typecompany blog

Dev: one capability, used by three customers.Dev: one capability, many customerscapabilitycustomercustomercustomerDelta: one customer, served across three capabilities.Delta: one customer, many capabilitiescapabilitycapabilitycapabilitycustomer
The product engineer's scope is a component used by every customer; the forward deployed engineer's scope is whatever one customer needs, across components.

In our reading, the customer’s problem sets your scope, not a component you own. Picture an illustrative week at a regional insurer.

  • Monday, the nightly policy extract from their mainframe arrives with a new column, and your pipeline rejects it.
  • Tuesday, claims adjusters must see only their own region’s claims, so you add row-level permissions keyed on each adjuster’s region.
  • Wednesday, your own product’s file connector times out on large files. You pin down the size, write a repro and hand it to core engineering rather than patching around it.
  • Thursday, you watch the first adjuster work a live claim in the new tool, then walk the claims director through it.

None of that is one component, and all of it is yours. Sierra’s Natalie Meurer, speaking at AI Engineer World’s Fair 2026, said that across versions of the role as different as DevOps, enablement, custom solutions and data integration, the one constant is that every forward deployed engineer is accountable to the customer. Source 7The 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 So your stories should show you on the hook: the customer called you, not a support queue, and you fixed it in their environment.

What a week contains, from people who did it

Asked how much of the job is not coding, say: “The FDEs I’ve read describing their weeks put building and running software at the center. The rest is where the scope gets decided, and I want to be in that room, because otherwise I build the wrong thing well.” Handed a vague brief, say: “Before I plan all of it, I’d build the thinnest version against their real data this week, because that’s where the brick wall shows up, then re-scope with them.” That thin first version is a walking skeleton.

The evidence comes from single voices, not a typical week:

  • Dyon, a Delta quoted on Palantir’s 2019 blog, spent “a couple of days working at the customer premises” most weeks, between stakeholder meetings and “monitoring, debugging, deploying, or configuring our software.” Source 1Dev versus Delta: Demystifying engineering roles at PalantirPublisherPalantir BlogSource typecompany blog
  • Het Trivedi, a Baseten FDE, wrote on Baseten’s blog, in a post last updated in May 2025, that over time his work split roughly 75% software engineering and model optimization, 15% technical consulting and solution design, and 10% customer relationship management. Source 8What I learned as a forward-deployed engineer working at an AI startupPublisherBasetenSource typecompany blog
  • Colin Jarvis, then OpenAI’s head of forward deployed engineering, was quoted in The Pragmatic Engineer, in August 2025, saying that what customers describe in scoping often does not match the data and system reality, so the team wants to “bias for moving fast, proving out any brick walls and then adjusting the scope.” Source 9What are Forward Deployed Engineers, and why are they so in demand? (Gergely Orosz)PublisherThe Pragmatic EngineerSource typenews report

The building permits case is built on that gap between the brief and the data.

How success is measured, company by company

The wording tells you what to say when asked how you would know an engagement worked.

  • Palantir’s 2019 Dev-versus-Delta blog post says Deltas “measure success in terms of impact on the customer’s goal.” Source 1Dev versus Delta: Demystifying engineering roles at PalantirPublisherPalantir BlogSource typecompany blog
  • OpenAI’s San Francisco posting says FDEs measure success “through production adoption, measurable workflow impact, and eval-driven feedback that changes product and model roadmaps.” Source 4Forward Deployed Engineer (FDE) - SFPublisherOpenAI CareersSource typecompany job posting
  • Microsoft’s Senior FDE posting, as of September 2026, says its founding team is “measured on adoption” and rules out the alternatives: “Not revenue, not renewals, not tickets closed.” Source 10Microsoft Careers: Senior Forward Deployed Engineer (WPD)PublisherMicrosoftSource typecompany job posting
  • Cursor’s EMEA FDE leadership posting, as of September 2026, says “the measure of success is what’s different at day 30, not what got shipped.” Source 11Regional Director, Forward Deployed EngineeringPublisherCursor (Ashby job board)Source typecompany job postingSource 12RVP, Forward Deployed Engineering - EMEAPublisherCursor (Ashby job board)Source typecompany job posting

None of these is lines of code. Shipping is the entry ticket, not the score: say what changed for the customer first, then what you shipped to cause it.

Databricks’ Senior FDE posting, as of September 2026, reads differently from Microsoft’s “not revenue”: it says “FDEs are billable and know how to complete projects according to specification.” Source 13Sr. Forward Deployed Engineer (FDE) - Financial Services (New York City)PublisherDatabricks (careers site / Greenhouse)Source typecompany job posting Where the posting says billable, your success story also needs the milestone you delivered to spec; reading a job posting takes that posting apart.

Answer in the terms of that company’s line, including in your first-month plan. For OpenAI, name the adoption signal and the eval: “the claims team sends new intake through the agent instead of re-keying it, and we rerun a set of real past claims on every prompt change to prove accuracy held.” For Microsoft, name who uses it and how you would see it: “the partner’s developers call the new API from their own build, and we see that in usage logs, not their status report.”

Use the posting's words, then interpret them

The blog posts above are history; the posting you applied to is current. In our reading, “measured on adoption” means your first-month plan starts with the team whose workflow hurts most, not the biggest logo.

Platform configuration or a bespoke build

The definitions above split on one question: do you make the vendor’s product work at the customer, or write new systems in their stack?

Platform work means configuring the vendor’s product until it solves this customer’s problem, as in Brian’s description of Palantir above. Cohere’s infrastructure FDE posting, as of September 2026, covers deploying its North product in private cloud and on-premises environments. Source 14Forward Deployed Engineer, Infrastructure Specialist (North America)PublisherCohere (Ashby job board)Source typecompany job posting

Bespoke work means new code in the customer’s environment, which Google’s posting describes as shipping “bespoke agentic solutions directly within the customer’s environment.” Source 6Forward Deployed Engineer IV, GenAI, Google Cloud — Google CareersPublisherGoogleSource typecompany job posting

To tell which you are in, read the verbs next to the product name. “Deploy”, “configure” or “implement” next to the vendor’s own product, as with North at Cohere, points to platform work. “Bespoke”, “custom” or “in the customer’s environment” points to a build. When a posting says both, ask the recruiter: “In the first month with a new customer, how much of the work is configuring your product, and how much is new code in the customer’s own repository?”

In our view, platform roles reward integration and adoption stories, and bespoke roles reward design and ownership of code in someone else’s production. Google’s GenAI FDE III posting lists evaluation pipelines and observability frameworks for agentic systems. Source 15Forward Deployed Engineer III, Generative AI, Google Cloud — Google CareersPublisherGoogleSource typecompany job posting

Platform roleBespoke role
Ships configuration, data integrations and deployments of the vendor’s productShips agents, MCP servers, evaluation pipelines and observability for agentic systems
Lead with integration and adoptionLead with design and code you owned

Some employers, and one former Palantir lead, want FDEs to lean toward work that ends up in the product.

A Cohere FDE wrote on Cohere’s blog, in August 2026, that when the product falls short for a customer, Cohere’s FDEs can inspect or modify the underlying code and work with core engineering on a fix that can be generalized across customers. Source 19FDEs: Building Capacity, Not DependencyPublisherCohereSource typecompany blog In our view, the failure mode is the opposite: an engineer who ships whatever the customer asks, in code only that customer runs, and stays because nobody else can maintain it.

In the interview, name one part of your last build that became reusable, one part you kept custom on purpose, and one request you declined because the product should do it. Prepare for questions about patterns you saw across customers, field feedback that changed a roadmap and when a customer does not need an embedded engineer at all. Then ask the hiring manager: “Of what your FDEs built last year, what ended up in the product?”

What the job is not

A Delta quoted in Palantir’s Dev-versus-Delta post says consultants generally create “a one-time analysis, recommendation, or solution to a specific problem,” while Palantir works with its customers to build “a long-term solution.” Source 1Dev versus Delta: Demystifying engineering roles at PalantirPublisherPalantir BlogSource typecompany blog Cursor’s posting is blunter: “FDEs don’t advise, they embed directly with customer engineering teams.” Source 11Regional Director, Forward Deployed EngineeringPublisherCursor (Ashby job board)Source typecompany job postingSource 12RVP, Forward Deployed Engineering - EMEAPublisherCursor (Ashby job board)Source typecompany job posting

If your best stories end with a recommendation someone else built, you have shown the discovery half of the job; show the build half too.

How close the role sits to the deal varies by employer: Palantir’s careers page places Deltas in Business Development, while a Baseten FDE manager posting says “FDE at Baseten is not a sales function.” Source 2Palantir Careers | Students and Early TalentPublisherPalantir TechnologiesSource typecompany hiring pageSource 20Engineering Manager, Forward Deployed Engineering (LLM) @ BasetenPublisherBaseten (Ashby)Source typecompany job posting Telling FDE from its neighbors shows which way yours leans.

Now say the three-sentence definition for your target posting, and name whether the role configures a platform or builds bespoke systems. Then answer the question about working inside a customer’s environment out loud, naming what you built and where it ran.

Keep going. That question is free, with an illustrative answer to compare yours against, and every one of the site’s 180 practice questions shows its framework free. Pro opens the illustrative answers to your first-month plan and the pattern you noticed across customers, the two stories where this lesson’s success measures and product lean get tested.

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 engineerGlossaryForward deployed software engineerPalantir’s title for its FDE role, called Delta internally; OpenAI and EY also post FDSE titles, each with its own duties.More on Forward deployed software 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 ProtocolGlossaryAgentA system in which a model chooses steps and tool calls to complete a task, within limits the design sets.More on Agent