What is a forward deployed engineer? Role, pay and Palantir origin
A forward deployed engineer (FDE) is a software engineer who works directly inside a customer's organization, building and adapting a vendor's product until it solves that customer's real problem. The role started at Palantir, and AI vendors such as OpenAI and Anthropic now hire for it under the same name. Think of it as a software engineer with a customer, a deadline and a production system to ship into.
What a forward deployed engineer does
The work sits between product engineering and consulting. An FDE does not maintain the vendor's core codebase. Instead, they take the product to a client, find out what the client's systems and data actually look like, and write the code that connects the two.
Job postings from AI companies describe a consistent set of duties:
- Embedding with strategic customers to understand technical needs and scope the project.
- Building working applications, integrations and agents inside the customer's environment.
- Coding alongside the customer's own engineers and handing over documentation.
- Reporting repeated problems back to the product and engineering teams so they can become features.
The last point matters commercially. A good FDE team turns one-off customer work into reusable product. A weak one becomes a custom-development shop.
Where the role came from: Palantir
Palantir is widely credited with creating the model. Its own engineering blog describes a split between product developers ("Devs") and engineers who deploy its platforms at customers ("Deltas"). See Palantir's posts Dev versus Delta and Who Wants to be a Delta?.
The idea behind it: Palantir sold software into government agencies and large companies with messy, sensitive data. Selling a license was not enough, so it put engineers on site to make the software work. The title "forward deployed" borrows from military language for units stationed close to the front.
How AI companies adopted it
Foundation model companies face a similar problem. A model API is easy to try and hard to turn into a dependable business process. Enterprise customers need evaluation, integration with internal data and guardrails, and many do not have the in-house skills to do that quickly.
That is why OpenAI and Anthropic have both advertised FDE roles. Two postings that are still online, both listed as more than six months old when we checked in October 2026, show what the job involves:
- An OpenAI posting for a Forward Deployed Software Engineer in New York describes embedding with strategic customers, building full-stack solutions iteratively and giving product feedback internally. It lists travel of up to 50%.
- An Anthropic posting for a Forward Deployed Engineer, Applied AI describes building production applications with Claude models inside customer systems, including MCP servers, sub-agents and agent skills, plus sharing deployment patterns with product teams.
The Anthropic listing is marked as no longer accepting applications, and the OpenAI page gives no closing date. Both are hosted on venture investors' job boards, which mirror the companies' postings. Check each employer's own careers page for current openings, since these roles open and close often.
Forward deployed engineer vs software engineer
The difference is who the engineer is accountable to and where the code runs.
- Software engineer (product): builds features for many customers, owns a codebase, and measures success by product quality and adoption.
- Forward deployed engineer: builds for one customer at a time, often in that customer's environment, and measures success by whether the customer's deployment works and expands.
FDEs write real production code, but they also run workshops, scope projects and explain trade-offs to non-technical stakeholders. The skill mix leans toward communication and speed as much as depth in one stack.
FDE vs solutions engineer vs sales engineer
The titles overlap, and companies use them inconsistently, so check the posting rather than the title.
- Sales engineer: supports the sales process with demos, proofs of concept and technical answers. The goal is to close the deal.
- Solutions engineer: often covers pre-sales design and post-sales onboarding, usually with lighter custom code.
- Forward deployed engineer: typically starts after the deal and writes substantial custom software to get the product live.
What FDEs are paid in the US
Pay ranges below come from the same two older job postings, not salary surveys, so treat them as past examples rather than current offers:
- OpenAI's New York forward deployed software engineer posting lists USD 200,000 to 310,000 a year plus equity.
- Anthropic's New York applied AI posting lists USD 200,000 to 300,000 a year in annual base salary.
Ranges differ by city, seniority and team, so treat them as examples rather than a market average.
Why AI deployments are services-heavy
The FDE model has a business consequence. Every engineer embedded at a customer is a salaried cost that does not scale like software. If a vendor sells a subscription but needs weeks of custom engineering per account, gross margin looks more like a services firm than a software company.
Investors and founders watch for a few signals:
- Whether deployment work is billed separately or bundled into the subscription.
- Whether patterns from early customers become product features, shrinking the engineering needed for the next customer.
- How many customers each FDE can support.
Our coverage of this tension includes "The FDE Trap" and a report on how enterprise AI adoption has stalled for some deployments, both in the AI Business section. For related background, see the explainers on AI wrappers and AI neolabs, and the Anthropic funding history. More company coverage is in AI and Startups.
Who suits the role
Postings tend to ask for several years of experience in a technical, customer-facing job, strong programming skills (Python is common in the AI listings) and comfort with ambiguity. Anthropic's posting asks for 3+ years; OpenAI's New York posting asks for 7+ years of full-stack experience. Former founders are often named as a good fit, because the job rewards ownership and fast decisions.
Travel can be significant, and some roles require time at customer sites. That is a lifestyle trade-off worth weighing alongside the pay.
Frequently asked questions
What is a forward deployed engineer in simple terms?
It is a software engineer who works at or with a customer to make a vendor's product fit that customer's systems. They write code, run the rollout and feed lessons back to the product team.
Is a forward deployed engineer the same as a software engineer?
No. A product software engineer builds features for all customers, while an FDE builds and deploys for one customer at a time. Both write production code, but FDEs spend more time with customers.
Did Palantir invent the forward deployed engineer?
Palantir is widely credited with originating the model, and its engineering blog describes Deltas as the engineers who deploy its platforms at customers. Many AI companies now use the title for a similar role.
How much does a forward deployed engineer make?
Two older US postings for New York roles, from OpenAI and Anthropic, listed ranges of roughly USD 200,000 to 310,000 a year, with OpenAI adding equity. Both were more than six months old in October 2026. Ranges vary by location and level.
Why are AI companies hiring forward deployed engineers?
Enterprise customers often need help integrating models with their data and processes before AI delivers value. FDEs close that gap, and their feedback shapes the product.