How to hire a forward deployed engineer

To hire a forward deployed engineer (FDE), make five decisions in order. Confirm the bottleneck is deployment work: code written inside a customer’s systems before your product delivers value. Choose how to staff it: a full-time hire, a contractor, a transfer of a product engineer, or someone you develop. Hire for engineering depth and customer judgment together; all four employers’ postings reviewed here ask for both. Test scoping and stakeholder work in the interview loop, not only code. Then set pay from posted ranges, such as $135,000 to $200,000 base in Palantir’s New York posting and $185,000 to $325,000 in an OpenAI San Francisco posting, before equity, as listed on October 22, 2026.^4,2

Do you need a forward deployed engineer, or a different hire?

An FDE is a software engineer whose job is getting your product into production inside a specific customer’s environment. OpenAI’s posting describes coding side by side with customers’ technical teams on the customer’s own infrastructure.²2OpenAI, “Forward Deployed Software Engineer - SF,” 15 November 2025. That is a different job from solutions engineering. The Bureau of Labor Statistics describes sales engineers as people who sell products or services “that require technical expertise,”⁷7U.S. Bureau of Labor Statistics, “Sales Engineers: Occupational Outlook Handbook.” and OpenAI’s posting names Solutions Engineering and Solutions Architects as separate functions working the same accounts.² One role helps win the deal. The other owns what happens after it.

Signals that you need an FDE:

• Pilots stall after signature. The contract closes, but integration with the customer’s data, identity systems and legacy software never reaches production.
• Product engineers are doing one customer’s work. Roadmap time goes to one account’s connectors and edge cases.
• Enterprise deals carry a scoped build. Anthropic’s posting for an FDE manager describes qualifying engagements, scoping the work and informing statements of work during pre-sales.⁸8Anthropic, “Manager, Forward Deployed Engineering,” 5 February 2026. If your deals include technical delivery, someone has to own it.
• Each deployment teaches you something the product needs. OpenAI’s FDEs feed field findings back to its research and product teams.⁵5OpenAI, “Forward Deployed Engineer (FDE) - SF,” 11 September 2026.

Signals that you do not:

• Customers adopt the product self-serve, or onboarding is configuration rather than code.
• Contract values are too small to carry an engineer’s time. The posted base pay below is only part of what an FDE costs.
• The real problem is winning deals: demos, security questionnaires, technical objections. That is a solutions engineering hire.
• The fix is product work every customer needs. Build it once with product engineers.

Whether the model suits your margins is covered in our analysis of when the forward deployed engineering model pays off.

Should you hire, contract, transfer or develop an FDE?

There are four ways to put an FDE in front of a customer. The deciding questions are how long the need will last and whether field learning has to stay in your company. The table is our qualitative assessment, not measured data; the Torchlane row is limited to what its hire page says.

* Our top pick. We have a business connection with Torchlane.

Contracting. An FDE embedded in your repositories, standups and customer sessions raises a real classification question. The IRS says the keys to worker status are the entire relationship and the extent of your right to direct and control the worker, and that classifying an employee as an independent contractor with no reasonable basis can leave you liable for employment taxes.⁹9Internal Revenue Service, “Independent contractor (self-employed) or employee?” Check classification with employment counsel before a contractor works inside your team.

Developing someone. Employers that hire less experienced FDEs surround them with support. Palantir asks for at least one year of post-college experience and describes small teams and mentorship.⁴4Palantir Technologies, “Forward Deployed Software Engineer (New York).” Salesforce says its FDEs work alongside senior FDEs and a Deployment Strategist in a pod, with structured mentorship.³3Salesforce, “Forward Deployed Engineer (FDE) (Mid/Senior Level),” 30 September 2026. Without that senior layer, the customer pays for the learning curve.

Torchlane. Torchlane is also our top pick for the develop route. Its AI FDE training has two starting points, technical experience and business experience, so an existing employee can build AI implementation skills on top of the expertise they already have. Torchlane says a beginner should not treat it as a substitute for the engineering experience a senior FDE role requires.¹⁰10Torchlane, “AI FDE training: forward deployed AI engineering.” That matches the point above: pair anyone you develop with a senior FDE or tech lead.

We have a business connection with Torchlane.

Transferring a product engineer. They bring product knowledge no outside hire has, but may lack the customer half, which Salesforce calls “a requirement, not a nice-to-have.”³ Test it first by having them lead one customer scoping session.

What should the forward deployed engineer profile require?

The postings reviewed overlap heavily: strong engineering, customer-facing delivery and, in three of the four, hands-on work with large language models (LLMs). They differ on seniority and travel. Four employers’ postings, checked October 22, 2026:

The engineering bar does not drop because the job faces customers: each posting asks for strong coding, and Salesforce, OpenAI and Anthropic ask for production experience explicitly.^4,3,5,66Anthropic, “Staff Forward Deployed Engineer,” 14 October 2026. Where LLM experience is asked for, it is defined practically: Anthropic says side projects and prototypes count, and Salesforce wants candidates who can explain why a prompt failed and what they would change.^6,3 And travel runs from about 25% to 50%, so state it plainly in your posting.

For a copy-ready posting, see our forward deployed engineer job description template.

Where do forward deployed engineer candidates come from?

Lightcast, a labor market data company, counted about 200 US FDE job postings in 2024, about 1,200 in 2025 and more than 5,200 in the first seven months of 2026.¹¹11Lightcast, “Q: What is a Forward Deployed Engineer? A: The fastest growing AI job,” 15 September 2026. Postings measure openings advertised, not people hired, but they show how many employers compete for the same candidates, so a search limited to people who already hold the title competes with all of them. Adjacent roles widen the pool, and the postings name several:

• Consultants and professional services engineers who have delivered technical projects for clients. Salesforce counts consulting, professional services and forward deployed engineering as qualifying backgrounds.³
• Tech leads with consulting or product experience, and former technical founders, both named in Anthropic’s staff posting.⁶
• Early startup engineers who built a product from scratch, which OpenAI lists as a plus.²
• Solutions architects and solutions engineers who still write production code. Anthropic’s manager posting counts solutions architecture experience.⁸ Not every solutions engineer still writes production code, so test it.

Each source tends to bring one half of the profile: consultants the customer judgment, product engineers the code. Weight the loop toward the half the candidate is missing.

Torchlane. Our top pick for employers who need forward deployed engineers is Torchlane. Share a hiring brief with the problem, the experience you need and the evidence you want to inspect, and Torchlane discusses fit. Introductions depend on a candidate’s background, evidence, fit and availability; an inquiry does not guarantee one, Torchlane does not publish a live candidate inventory, and you remain responsible for assessment, terms and selection.¹1Torchlane, “Hire AI engineers and forward deployed AI talent.” Torchlane says program completion alone does not establish senior engineering experience, and that its assessment is not accreditation or a guarantee of employment.^1,1212Torchlane, “AI skills assessment and credentials.”

We have a business connection with Torchlane.

How should the interview loop differ from a standard engineering loop?

A standard software loop tests coding and system design. An FDE loop adds what the postings stress: scoping, stakeholder judgment and delivery inside someone else’s environment.

1. Production walkthrough. The candidate walks through a system they shipped and operated, including what broke. Salesforce asks candidates for a production system they would be proud to walk its team through.³
2. Scoping exercise. Give a business problem, not a specification, and ask for a technical plan with risks and sequencing. Anthropic’s staff posting looks for someone who can walk into discovery with only a business problem and leave with a defensible technical strategy.⁶ OpenAI’s forward deployed software engineer posting includes preparing detailed scopes of work.²
3. Integration coding. Code against an unfamiliar API or messy data, ideally with an LLM in the loop, then diagnose a failure. This is closer to the job than a puzzle.
4. Stakeholder session. A role-play with a skeptical customer executive or security reviewer. OpenAI’s FDEs work closely with its governance, risk and compliance and security teams, and Anthropic expects staff FDEs to anticipate customers’ security and procurement gates.^5,6

Score each stage separately. A candidate who excels at code and struggles with customers is a strong product engineer, not an FDE.

How much should you pay a forward deployed engineer?

A pay-transparency range is one employer’s estimate for one posting at one time. New York’s law, for example, requires employers with four or more employees to post a range believed to be accurate at the time of posting.¹³13New York State Department of Labor, “Pay Transparency.” It is not a record of what hires accepted. Figures below were checked on October 22, 2026.

How to use these numbers:

• Match level and location, not title. Salesforce’s single range covers mid and senior levels across many US locations, and the posting says pay depends on factors such as location and level.³
• Do not average across postings. Levels, cities and pay mixes differ, and none values its equity.
• Decide what you compete on. If you cannot match a lab’s base, compete on scope, ownership and equity, and say so in the posting.
• Budget travel separately. The postings state about 25% to 50%.

What should a new FDE deliver in the first 90 days?

We found no public benchmark for how fast an FDE ramps, so define success before the person starts, in outcomes rather than hours. OpenAI’s posting says its FDEs measure success through production adoption, measurable workflow impact and eval-driven feedback that changes product and model roadmaps.⁵ Anthropic asks staff FDEs to hand over systems the customer can run and to codify repeatable deployment patterns.⁶ Salesforce has its FDEs build dashboards that track deployment health and business outcomes.³

A workable 90-day plan, our recommendation rather than an industry standard:

• Days 1 to 30. Learn the product, shadow a live account, and write one customer’s deployment plan: scope, risks and the customer’s success measure.
• Days 31 to 60. Ship a first component into that customer’s production environment, through their security review.
• Days 61 to 90. Customer users in the system, a handover their team can operate, and written field findings your product team has accepted or declined.

Track time to first production deployment, active customer use, and product changes from the field. If 90 days produce custom code only one customer can use and no product learning, the role is drifting into services work; our guide to designing a forward deployed engineering team covers how to prevent that.