A forward deployed engineer (FDE) is a software engineer who works inside a customer's team, scoping the problem, building the solution on the customer's real systems and staying until it runs in production. The role started at Palantir. AI companies like OpenAI, Anthropic and Clay now hire FDEs to turn their products into working customer outcomes.

Key points

• An FDE writes code for one customer's problem, while a product engineer writes code for every customer at once.

• A GTM-focused FDE does the same job inside sales, marketing and RevOps systems: CRM, enrichment, routing and AI agents.

• Sales engineers help close the deal. FDEs make the product work after it closes.

• Most paths into the role start from GTM engineering, solutions work or consulting, plus real production builds.

• Companies that need this help without a full-time hire can bring in a fractional engineer for a defined scope.

If you are mapping a career move, start with the guide to becoming a GTM engineer. The FDE track is one of the clearest branches from it.

Where did the forward deployed engineer role come from?

Palantir popularized it. Semafor's tech editor Reed Albergotti wrote that "the concept for the role was popularized by Palantir," and that AI startups are now chasing the same skill set.

Palantir's own engineers call the role "Delta." In a Palantir blog post titled Who Wants to be a Delta?, a forward deployed software engineer describes how a typical engineer's "scope of influence is just one leg in what I think of as a relay race." Deltas run the whole race. They "engage in initial customer discussions to explore, identify, and scope out the customer's business problems," and they are "there with the users when the applications go live."

That last line is the heart of it. An FDE is measured by whether the customer uses the thing, not by whether the code shipped.

Why are AI companies hiring forward deployed engineers now?

Because AI products rarely work out of the box. A model has to be wired into each customer's data, tools and approval rules before it does anything useful, and someone has to do that wiring.

PYMNTS reported that monthly job listings for forward-deployed engineers increased by more than 800% between January and September 2025, citing the Financial Times. The same article says the role "has roots in Palantir" and that companies "including OpenAI, Anthropic and Cohere are expanding teams of forward-deployed engineers."

The job posts show what the work looks like at each company:

• OpenAI. Its Forward Deployed Software Engineer posting asks engineers to "embed deeply with strategic customers to understand their business challenges and technical requirements in detail." It lists a range of $185K to $325K plus equity and asks for 7+ years of full stack experience.

• Anthropic. A Forward Deployed Engineer, Applied AI posting for Paris says the FDE will "deliver technical artifacts for customers like MCP servers, sub-agents, and agent skills." It expects frequent travel, "25-50%," and lists €205,000 to €220,000.

• Clay. A Forward Deployed GTME posting describes "a new Services team to help Clay's largest customers design, build, and launch high-impact workflows," at $130,000 to $180,000 a year.

Clay's version is the GTM one. The title literally merges the two roles.

What is a GTM-focused forward deployed engineer?

A GTM-focused FDE is an FDE whose customer problems live in revenue systems. Instead of data pipelines for a government agency, the builds are lead routing, enrichment, outbound sequences, CRM hygiene and AI agents that draft or qualify.

The Clay posting is a useful map of the scope. The engineer is asked to:

1. Run discovery "with a range of personas" across RevOps, marketing ops, sales and customer success.

2. Build "complex workflows in Clay (lists, enrichment, automations, AI prompts, audiences, and reporting) and document them."

3. Map the process by "drawing on light business process consulting skills to map existing workflows."

4. Own delivery from kickoff through go-live, including timelines and risk.

The skill base overlaps heavily with GTM engineering. If that term is new, what a GTM engineer does covers the core role. The difference is who you serve: a GTM engineer usually builds for their own company, while a GTM FDE builds for a vendor's customers, one account at a time.

How does an FDE differ from a solutions engineer, sales engineer, consultant and GTM engineer?

The shortest answer: ask who pays the engineer, and when in the customer's journey they show up. FDEs arrive after the sale and stay until the build is live.

Salesforce describes a sales engineer as someone "acting as a bridge between a customer or prospect's technical resources and their company's sales team." That bridge role is mostly pre-sale. An FDE is past the bridge and building on the other side.

Titles blur in practice, so read the job post rather than the title. If a role is mostly demos, it is sales engineering whatever it is called. For a related split inside one company, see GTM engineer vs RevOps vs sales ops.

A habit that separates good forward deployed engineers: they log the work as they go, so every prompt, source and decision is captured while it is fresh.

What does a week look like for a GTM forward deployed engineer?

Mostly building, with structured time on calls. Here is an illustrative week for an FDE assigned to one large customer, based on the responsibilities in the job posts above. It is a composite, not a real schedule.

Monday: discovery and scoping

• Interview the RevOps lead and two SDR managers about how inbound leads get routed today.

• Write a one-page scope: the problem, the success metric, what is out of scope.

Tuesday and Wednesday: build

• Connect the vendor's product to the customer's CRM and enrichment sources.

• Build the first version of the workflow in a sandbox, using real sample records.

• Add a human approval step anywhere the workflow sends an email or changes an owner.

Thursday: working session

• Walk the customer's team through the build on a shared screen.

• Collect edge cases they spot, such as duplicate accounts or region rules nobody wrote down.

Friday: harden and write it up

• Fix the edge cases and document the workflow so the customer's team can maintain it.

• Send product feedback to the vendor's own engineers, flagging anything that needed a workaround.

The Friday handoff matters. Clay's posting asks for workflows that are documented, and Anthropic's asks FDEs to "identify and codify repeatable deployment patterns." A build only the FDE understands is a liability.

What skills does a forward deployed engineer need?

Three groups: engineering, GTM systems, and customer work. Most candidates are strong in one and need to build the other two.

Engineering

• A working language. Python appears in the Anthropic posting as a strong requirement, with TypeScript as a plus.

• APIs and webhooks. Most GTM builds move records between tools that were never designed to talk.

• LLM work. Anthropic's posting asks for "production experience with LLMs" covering prompting, agents and evaluation.

• Data basics. Relational databases appear in the OpenAI posting.

GTM systems

• CRM data models. Accounts, contacts, opportunities and the fields that break routing.

• Enrichment and outbound tools. Clay's posting names CRMs, marketing automation, sales engagement tools and enrichment platforms.

• Approval design. Knowing which steps an AI can take alone and which need a person. The L1 to L4 AI automation maturity ladder is one way to set those rules.

Customer work

• Discovery. Asking the questions that surface the real problem.

• Scoping. Saying no to the extra request that would sink the timeline.

• Explaining. Talking to an SDR manager and a CTO in the same afternoon.

For a level-by-level breakdown, the GTM engineer skills matrix maps these across Foundation, Working and Advanced levels.

How do you become a forward deployed engineer in GTM?

Build real GTM workflows in production, then show them to a vendor that deploys into revenue teams. The job posts ask for customer-facing experience as much as code, so you need proof of both.

1. Pick one stack and go deep. A CRM, one enrichment tool, one automation tool and one model provider. The GTM engineer tool stack guide lists the usual choices.

2. Ship three workflows that run every day. For example lead routing, account research and a reply drafter with an approval step. Demos do not count.

3. Document each one like a handoff. Trigger, inputs, logic, failure cases and who approves what.

4. Get customer-facing reps. Implementation, solutions or consulting work all count. Clay's posting asks for 3 to 7+ years in exactly that kind of role.

5. Write up what broke. A short post about an edge case you fixed is better proof than a certificate.

6. Target vendors whose product you already know. Your fastest interview is with a company whose tool you have already shipped on.

7. Prepare for scoping questions. Expect to be handed a vague customer problem and asked to turn it into a plan.

The interview side gets its own breakdown in the GTM engineer interview audit.

What are the common mistakes people make with the FDE path?

The biggest one is treating it as a sales role with a technical title. FDEs build, and the interviews test that.

1. Applying with only demos. Hiring teams want workflows that ran on real data for weeks.

2. Ignoring the travel. Anthropic's posting lists 25 to 50 percent travel. Ask before you accept.

3. Building without documentation. The handoff is part of the deliverable.

4. Skipping approval design. An AI step that emails prospects with no review is the fastest way to lose a customer's trust.

5. Chasing the title over the scope. Some "FDE" posts are support roles. Read the responsibilities, not the heading.

6. Staying on one customer too long. The value of the role is the pattern library you build across accounts. If every engagement looks the same, write the pattern down and move on.

Whoever does the work, the handoff decides whether it lasts. A one-page contract per agent covers the boundary, the escalation trigger, the owner and the rollback steps.

How can a company get FDE-style help without hiring one?

Hire the work, not the headcount. A vendor's FDE only builds on that vendor's product, and a full-time hire is a large commitment for a few months of build work.

The usual alternative is a fractional engineer: someone who embeds with your GTM team for part of the month, works across all your tools, and leaves behind documented workflows. The guide to hiring a fractional AI engineer covers scoping, pricing models and what to ask before you sign.

When each option fits

To put a number on that last row, use the worked example from the AI Workflow Install page: a 12-person team where each person spends 8 hours a week on manual work, at a $60 loaded rate, costs $299,520 a year. In that example, one workflow accounts for $149,760 of it. That is the kind of figure that tells you whether an embedded engineer, fractional or full-time, pays for itself.

FAQ

What does FDE stand for?
Forward deployed engineer. Palantir also uses the title forward deployed software engineer, and calls the role "Delta" internally.

Is a forward deployed engineer a sales role?
No. FDEs usually work after the sale and write production code. Sales engineers are the pre-sale technical role.

How much do forward deployed engineers earn?
It depends on company and level. OpenAI's posting lists $185K to $325K plus equity, and Clay's Forward Deployed GTME posting lists $130,000 to $180,000.

Do I need a computer science degree to become an FDE?
Not always. Anthropic's posting accepts a bachelor's degree "or equivalent experience," and GTM-focused roles weigh shipped workflows heavily.

Is a GTM engineer the same as a forward deployed engineer?
They share most skills. A GTM engineer builds for their own company, while a GTM FDE builds for a vendor's customers.

Can a small company hire an FDE?
Most small companies do better with a fractional engineer for a defined scope. Full-time FDEs are typically employed by vendors to serve their largest accounts.

Talk through your FDE or GTM engineering plans

If you are weighing a move into forward deployed work, or deciding whether your team needs embedded engineering help, book an intro call and we can work out where to start. Hiring managers who want to cost the manual work first can book a free Leak Call.

About the author: Kevin Payne is an AI engineer and GTM operator who installs AI workflows for B2B teams. He has 10+ years in B2B SaaS demand generation and RevOps across Techstars, a16z and YC portfolio companies, and his writing has appeared in HubSpot, GetResponse, Outbrain and more.