Skip to content

What is a forward deployed engineer?

A forward deployed engineer builds production software with a vendor’s customers, often inside their systems. Here is what that means stage by stage, where the title came from and why one title can describe quite different jobs.

By All Things FDE editorial teamPublished Updated

Contents. Each line links to its section.

Contents

Key points

  • A forward deployed engineer is a software engineer who works with a vendor's customers, often inside their systems, to scope, build and roll out production software, and who is expected to write production code, not only demonstrate or configure a product [1][2][3].
  • Palantir says it pioneered the role, and its own posting calls the role the original; First Round Review and The Pragmatic Engineer also credit Palantir. Its job board labels forward deployed software engineers "Delta" and deployment strategists "Echo" [4][5][3][6].
  • The title is not standardized: in postings checked on October 23, 2026 the title or a close variant sat in engineering at Ramp, Applied AI at Anthropic, Professional Services Operations at Databricks and go-to-market at Harvey [7][2][8][9].
  • O*NET, the US Department of Labor's occupational database, has no occupation with the title, so size estimates come from private posting analytics: Lightcast reports about 200 US postings in 2024, around 1,200 in 2025 and more than 5,200 from January to July 2026, without publishing how it counted [10][11].
  • Commentators such as Marty Cagan and The Pragmatic Engineer put the product feedback loop at the center of what separates the role from consulting, and OpenAI, Anthropic and Scale AI each ask FDEs for it in their postings [12][6][1][2][13].

A forward deployed engineer (FDE) is a software engineer who works for a technology vendor but spends much of the job with the vendor's customers, often inside their systems, building and shipping production software for them. The work runs from discovery and scoping through build, rollout and handoff [1][14], and well-run teams feed what they learn back into the product [12][6]. Palantir says it pioneered the role, and AI labs and enterprise software companies now hire for it under many titles [4][1][2].

Everything below comes from employer job postings read on October 23, 2026, one SEC filing, a search of the Department of Labor's O*NET database and named published analyses. A posting is one company's wording at one moment, so we say which company each claim comes from, and where nobody has published reliable numbers we say so. For how the role differs from neighboring jobs, see our comparisons of the forward deployed engineer and the solutions engineer and of the forward deployed engineer and the AI engineer.

What does a forward deployed engineer do?

The postings we read describe the same arc in different words. Set out as stages, with the employer wording behind each:

  1. Discovery. Learn the customer's workflow, systems and constraints. OpenAI says its FDEs own discovery, technical scoping, system design, build and production rollout [1]. Ramp's AI Solutions posting asks the engineer to understand current workflows, systems, data quality and adoption blockers [14].
  2. Scoping. Turn that into requirements and a plan. Anthropic's staff-level posting has the engineer lead discovery and advisory engagements that shape scope and technical strategy before a build starts [15]. Ramp's lists a system context diagram, an integration plan, a security model, an evaluation plan and an operational plan as artifacts the engineer maintains [14].
  3. Build. Write production code in or beside the customer's environment. Anthropic's London posting asks for production applications built with Claude models inside customer systems and for artifacts such as MCP servers, sub-agents and agent skills [2]. Scale AI's public-sector posting names backend systems and data integrations that connect customer environments to its platform [13].
  4. Rollout. Take it to production and get it used. OpenAI says it measures the work by production adoption, measurable workflow impact and evaluation-driven feedback [1].
  5. Handoff. Leave the customer something it can run. Anthropic's staff-level posting says the engineer hands over production systems customers can run, and Ramp's lists operational handoff [15][14].
  6. Feedback to product. Turn repeated custom work into something reusable. OpenAI, Anthropic and Scale AI each ask for this in their postings [1][2][13].

That last stage is what separates the job from pure delivery work. Marty Cagan of the Silicon Valley Product Group describes FDEs as technical people who embed with a customer to understand its environment and problems and what it takes to deliver the outcomes the customer needs, and who work across several customers so that one product can serve them [12]. First Round Review puts it more briefly: the FDE builds the "last mile" so the product works in production and, unlike a solutions consultant or sales engineer, writes and debugs production code [3].

Where did the forward deployed engineer role come from?

From Palantir, by its own account and by two outside accounts. Its current New York posting calls the role the original and says the company pioneered the position by embedding engineers with customers [4]. First Round Review and The Pragmatic Engineer both credit Palantir as well [3][6]. Palantir's 2020 registration statement for its public listing refers to its forward-deployed engineers in a risk factor about customers renewing or expanding their use of its platforms. Its business description says its forward deployed engineers have travelled to bases in Afghanistan and factories in the industrial Midwest to deploy its platforms, and that time in the field feeds continuous improvement of those platforms [16].

Palantir's job board labels the teams: the Forward Deployed Software Engineer posting sits under the team name "Delta" and the Deployment Strategist posting under "Echo" [4][5]. The Pragmatic Engineer reports that Palantir created the role in the early 2010s, called it Delta, and until about 2016 employed more FDEs than regular software engineers, after which many moved back to core engineering when it launched Foundry [6]. First Round Review dates the model more loosely, to nearly two decades ago, and ties the word "deployed" to engineers spending their days onsite with customers [3]. We found no account from Palantir of why it chose the words "forward deployed", so we do not offer one.

Cagan calls the association with Palantir unfortunate because the model applies well beyond Palantir's hardest customers. In his framing, Palantir applied the product model to custom work: it bets on solving the customer's problem instead of building a specification the client dictates [12].

What are the other titles for the same work?

The title is not standardized. The table lists eight postings from employers' own job boards, checked on October 23, 2026. "Where the posting sits" is the department or team label the employer's job board shows. Pay is the range each posting publishes, in the basis it states. Amounts are as published on October 23, 2026 and read as annual; Databricks and Harvey print no period in the posting text. Each is one posting at one time and can span levels and locations, so none is a market rate.

Table 1. What are the other titles for the same work?
Title and employerWhere the posting sitsWhat it emphasizesPublished pay
Forward Deployed Software Engineer, Palantir, New York [4]Team "Delta"Embeds with customers and builds on Palantir's platforms; 1+ years after college; travel up to 25%$135,000 to $200,000 estimated salary; stock units and sign-on bonus may be added
Deployment Strategist, Palantir, New York [5]Team "Echo"Onsite discovery, workflows, training and presentations; works with FDEs on data integration; travel 25% to 75%$110,000 to $170,000 estimated salary; same exclusions
Forward Deployed Engineer, OpenAI, San Francisco [1]Forward Deployed EngineeringOwns discovery to production rollout and writes production code; 5+ years; travel up to 50%$185,000 to $300,000 pay range, plus equity
Forward Deployed Engineer, Anthropic, London [2]Applied AIBuilds production applications on Claude inside customer systems; 4+ years; travel 25% to 50%£225,000 to £255,000 annual salary (the posting says ranges for sales roles are on-target earnings)
AI Engineer, Forward Deployed Engineering (AI FDE), Databricks, US [8]Professional Services OperationsProfessional-services engagements building GenAI applications; heading says all levels, but the text says it is not intended for entry-level applicants$182,000 to $250,208 pay range; bonus and equity may be added
Software Engineer, Forward Deployed, Ramp, New York [7]EngineeringAgent-first, customer-facing team that works with Sales and Customer Success; 3+ years preferred$168,000 to $285,000 pay range, plus equity
Agent Deployment Engineer, Decagon, San Francisco [17]Customer TeamEnd-to-end execution of AI agent builds for strategic customers$175,000 to $230,000 pay range, plus equity
Forward Deployed Finance Engineer, Harvey, New York [9]Go to MarketFormer investment professionals who lead discovery, demos and pilots alongside account executives$254,600 to $350,000 on-target earnings (75/25 split), plus equity; not base salary

Other names for the same or neighboring work include forward deployed software engineer (FDSE), the term Palantir and Scale AI use in their titles and OpenAI posts alongside "Forward Deployed Engineer" and "Deployment Lead, FDE" [4][13][18]. First Round Review also mentions legal engineer, the name Ironclad used for its equivalent team, and notes that some founders who ask for an FDE turn out to need a post-sales role such as an implementations consultant [3]. Cagan notes that FDEs are usually engineers but increasingly product managers or designers [12].

Three patterns stand out. First, the same or a closely related title sits in different parts of the company: a dedicated FDE group at OpenAI, engineering at Ramp, Applied AI at Anthropic, professional services at Databricks, a customer team at Decagon and go-to-market at Harvey [1][7][2][8][17][9]. That matters because the department is one clue to whether field work reaches the product team or stays with the account, though a label is not proof. Second, experience asked runs from 1+ years after college at Palantir to 5+ years at OpenAI [4][1]. Third, several employers split the work. Palantir pairs FDEs with Deployment Strategists, and Ramp's AI Solutions posting has the engineer co-lead customer engagements with a strategist who owns business discovery and the return-on-investment case [5][14].

The a16z partner Joe Schmidt adds a caution about the label. He wrote in June 2025 that at AI application companies the job of guiding and managing AI agents is often filled by a professional-services employee, sometimes rebranded as a forward deployed engineer or an implementation or solutions specialist [19]. Read the duties, not the title.

What is a forward deployed engineer not?

Not a configuration or demo job. First Round Review contrasts the FDE with a solutions consultant or sales engineer by saying the FDE writes and debugs production code, and quotes one interviewee saying that configuration and implementation are not FDE work [3]. The Pragmatic Engineer reports that at OpenAI, solutions architects rarely write code on customer infrastructure and usually build proofs of concept with anonymized or offline data, while FDEs write code directly on customer infrastructure [6]. That describes one company's split, not a rule every employer follows.

Not consulting, when it works as designed. Cagan separates custom-solutions firms, which build what the client specifies for an agreed price, from the product model [12]. The Pragmatic Engineer says consultants typically make one-off recommendations while FDEs work with a customer over the long term and can add missing capabilities to the vendor's product [6]. Cagan also warns that a company with only FDEs would accumulate large bespoke solutions to maintain indefinitely [12]. If field work never reaches the product, the job looks like services work whatever its title.

Not a sales role, with exceptions. First Round Review quotes its head of talent saying that a founder who wants an FDE to have closed deals or run a set number of demos probably wants someone closer to sales, and describes a founder who wanted repeatable deployments and close work with sales and landed on an implementations lead [3]. Schmidt says it is rare to find implementation leads who want to carry a quota [19]. Yet Harvey's finance-engineer posting sits in go-to-market and has the engineer work with account executives through the pre-sales process [9]. Where a posting measures you on demos and deals, read it as closer to a sales role than an engineering one.

Not an O*NET occupation. A keyword search of the Department of Labor's O*NET database on October 23, 2026 returned no occupation titled forward deployed engineer; the results were loose matches such as software developers, sales engineers and engineers, all other [10]. So O*NET has no occupation-level entry for the title itself, and the size estimates we found come from posting analytics or company statements.

How big is demand, and what does the role pay?

Lightcast, a labor-market analytics firm, reports about 200 US postings for forward deployed engineers in 2024, around 1,200 in 2025 and more than 5,200 from January to July 2026 [11]. The page does not say how postings were counted or de-duplicated, and it describes demand as small next to established technology occupations [11]. Treat the direction as credible and the exact counts as approximate. Other growth percentages circulate: First Round Review cites an 800% rise in monthly listings between January and September 2025 and links a Financial Times article we could not open, so we do not rely on it [3].

On pay, the table's ranges are the most reliable figures we can offer, and they differ in basis. Palantir and Anthropic call theirs salary, and Palantir says its estimate excludes the value of any sign-on bonus, benefits and long-term incentives [4][2]. OpenAI, Ramp and Decagon list a range plus equity without saying more about its basis [1][7][17]. Databricks says its range is the expected salary for non-commissionable roles or on-target earnings for commissionable ones, and Harvey's is on-target earnings with a 75/25 split [8][9]. Scale AI's public-sector posting publishes base-salary ranges of $184,000 to $292,560 for San Francisco, New York and Seattle, $165,600 to $263,304 for Hawaii, Washington DC, Texas and Colorado, and $138,000 to $219,420 for St. Louis, and says each range may cover several career levels [13]. We do not average these into one number, because the postings differ in level, location and what they include.

How much coding and travel does the job involve?

Coding is central in the engineering-led postings. OpenAI asks for production-grade code across frontend and backend, Palantir for a strong coder, Scale AI for strong production coding in Python and TypeScript, and Anthropic for strong Python plus production experience with large language models [1][4][13][2]. Domain-expert roles differ: Harvey's asks for deal experience and treats Python or SQL as a plus [9].

Stated travel runs from up to 25% in Palantir's New York posting to up to 50% in OpenAI's, with Anthropic's London posting at 25% to 50% and Databricks asking for a customer visit about once every four to eight weeks [4][1][2][8]. Harvey says you may be asked to travel around two to three or more times a month, typically one to three days at a time [9]. Some employers give a percentage and others describe a pattern, so compare like with like before reading much into the gap.

Common questions

Is a forward deployed engineer just a consultant?

Not by design, though it can drift that way. Marty Cagan of the Silicon Valley Product Group says traditional custom-solutions firms build what the client specifies for an agreed price, while Palantir applied a product model to custom work. The Pragmatic Engineer says consultants typically make one-off recommendations, whereas FDEs work with a customer over the long term and can add missing capabilities to the product. The distinction holds only when field work feeds the product, so check where the posting reports and who owns that handoff.

Do forward deployed engineers write code?

In the engineering-led postings we read, yes. OpenAI’s asks for production-grade code across frontend and backend, Palantir’s for a strong coder, and Scale AI’s for strong production coding in Python and TypeScript. Exceptions exist: Palantir’s Deployment Strategist and Harvey’s finance engineer are built around domain and customer work, with programming a requirement or a plus rather than the core. A posting that never mentions production code may describe a different job.

Is a forward deployed software engineer the same as a forward deployed engineer?

In practice the terms describe the same function. Palantir uses forward deployed software engineer (FDSE) for the role it says it pioneered, and OpenAI’s job board lists both titles across different cities. Employers also vary the title by domain or level, for example Databricks’ AI FDE or Harvey’s forward deployed finance engineer, so read the duties and the reporting line rather than relying on the exact words.

How much do forward deployed engineers travel?

It depends on the employer. The postings we compared state up to 25% at Palantir, up to 50% at OpenAI, 25% to 50% at Anthropic’s London role, a customer visit about every four to eight weeks at Databricks, and around two to three or more trips a month at Harvey. Each is one posting at one time, and some are percentages while others describe a pattern.

How much experience do you need to be a forward deployed engineer?

The postings we read ask for 1+ years after college at Palantir, 3+ years preferred at Ramp, 4+ years in a technical customer-facing role at Anthropic’s London role, and 5+ years at OpenAI, at Scale AI’s public-sector role and at Decagon. Databricks says its AI FDE role is not intended for entry-level applicants. Experience asked varies by level and employer, so treat these as examples, not a standard.

Sources

  1. [1]OpenAI, “Forward Deployed Engineer (FDE) - SF”, 11 Sep 2026, accessed 23 Oct 2026. Back to text
  2. [2]Anthropic, “Forward Deployed Engineer (London, UK)”, 14 Sep 2026, accessed 23 Oct 2026. Back to text
  3. [3]First Round Review, “So You Want to Hire a Forward Deployed Engineer: How to Know If You Need One and How to Get the Role Right”, 24 Feb 2026, accessed 23 Oct 2026. Back to text
  4. [4]Palantir Technologies, “Forward Deployed Software Engineer (New York)”, accessed 23 Oct 2026. Back to text
  5. [5]Palantir Technologies, “Deployment Strategist (New York)”, accessed 23 Oct 2026. Back to text
  6. [6]The Pragmatic Engineer (Gergely Orosz), partly paywalled, “What are Forward Deployed Engineers, and why are they so in demand?”, 12 Aug 2025, accessed 23 Oct 2026. Back to text
  7. [7]Ramp, “Software Engineer, Forward Deployed (New York)”, 22 Oct 2026, accessed 23 Oct 2026. Back to text
  8. [8]Databricks, “AI Engineer - Forward Deployed Engineering (AI FDE)”, 20 Oct 2026, accessed 23 Oct 2026. Back to text
  9. [9]Harvey, “Forward Deployed Finance Engineer, Private Credit”, 6 Oct 2026, accessed 23 Oct 2026. Back to text
  10. [10]U.S. Department of Labor, O*NET OnLine, “O*NET OnLine keyword search: forward deployed engineer”, accessed 23 Oct 2026. Back to text
  11. [11]Lightcast, “Q: What is a Forward Deployed Engineer? A: The fastest growing AI job.”, 15 Sep 2026, accessed 23 Oct 2026. Back to text
  12. [12]Silicon Valley Product Group (Marty Cagan), “Forward Deployed Engineers”, 17 Sep 2025, accessed 23 Oct 2026. Back to text
  13. [13]Scale AI, “Forward Deployed Software Engineer, Public Sector”, 22 Oct 2026, accessed 23 Oct 2026. Back to text
  14. [14]Ramp, “Software Engineer, Forward Deployed AI Solutions”, 28 Jul 2026, accessed 23 Oct 2026. Back to text
  15. [15]Anthropic, “Staff Forward Deployed Engineer (New York City, San Francisco, Seattle)”, 14 Oct 2026, accessed 23 Oct 2026. Back to text
  16. [16]U.S. Securities and Exchange Commission (EDGAR), “Palantir Technologies Inc. Form S-1 Registration Statement”, 25 Aug 2020, accessed 23 Oct 2026. Back to text
  17. [17]Decagon, “Agent Deployment Engineer (San Francisco)”, 20 Jan 2026, accessed 23 Oct 2026. Back to text
  18. [18]OpenAI, “OpenAI Jobs”, accessed 23 Oct 2026. Back to text
  19. [19]Andreessen Horowitz (Joe Schmidt), “Trading Margin for Moat: Why the Forward Deployed Engineer Is the Hottest Job in Startups”, 4 Jun 2025, accessed 23 Oct 2026. Back to text

How this entry is made: each factual claim carries a numbered note to one of the sources above, and the date after each source is when the All Things FDE editorial team last checked it. Editorial standards.