How to hire a forward deployed engineer in 2026: roles, pay, vetting

A forward deployed engineer (FDE) is a software engineer who works inside one customer's business until an AI system or product runs on their real data and the team uses it every day. Monthly FDE job listings grew more than 800% between January and September 2025 (Indeed data for the Financial Times, via Fast Company), and OpenAI, Anthropic and AWS are all building FDE teams.
This guide is for founders and team leads deciding whether to hire one. It covers what FDEs actually do, why the role took off, what they cost and how to check a candidate before you commit.
  • What a forward deployed engineer is
  • What an FDE does, step by step
  • FDE vs solutions engineer, consultant and software engineer
  • Why FDEs are in demand in 2026
  • Do you need one?
  • What an FDE costs
  • Skills to look for
  • How to vet an FDE in 1 week
  • How to hire one this week

What a forward deployed engineer is

Palantir invented the role in the early 2010s and called it "Delta" (The Pragmatic Engineer). Palantir split its engineers in 2:
Role
Focus
Works on
Dev (product engineer)
1 capability, many customers
The core platform everyone uses
Delta (forward deployed engineer)
1 customer, many capabilities
Whatever that customer needs to get value from the platform
Until about 2016, Palantir had more FDEs than regular software engineers. Palantir describes the job as similar to a startup CTO's: own the problem, the code and the result, with no one to hand off to.
The definition that holds across companies today: an FDE writes production code on the customer's infrastructure and is accountable for the outcome. They split time between the customer and their own product team, and bring back what they learned so the product gets better for the next customer.

What a forward deployed engineer does

OpenAI's FDE team runs projects in 3 phases: scoping, validation and delivery (The Pragmatic Engineer). In practice, a good engagement has 5 steps.

1. Scope: sit with the people who do the work

The FDE spends the first days with the people the system is for: support agents, analysts, ops managers. They map what happens today, where the data lives, who approves what, and which step wastes the most hours. At OpenAI this means a couple of days onsite before any code.

2. Agree on how success is measured

Before building, the FDE and the customer agree on evals: a set of real cases with the right answers, and a score the system has to hit. Without this, every AI project ends in "it feels better" or "it feels worse," and nobody signs off.

3. Build inside the customer's systems

This is most of the calendar. Connect the CRM, the ERP, the shared drive full of PDFs. Handle permissions and the data that's wrong. Build the agent or the feature, then run it against the evals until it passes. Solutions architects rarely write code on customer infrastructure; FDEs do it daily.

4. Get people to actually use it

The technical part is often the shorter one. When OpenAI deployed GPT-4 at Morgan Stanley, the technical scaffolding took 6-8 weeks. Getting financial advisors to trust it took another 4 months of pilots, evals and changes. In the end, about 98% of advisors adopted it (Business Insider).

5. Feed it back into the product

What the FDE builds for one customer becomes a feature, a template or a playbook for the next. At OpenAI, an FDE project with a voice customer produced evals that went back to the research team and improved the Realtime API. This step is what separates an FDE from a consultant.
The tech took 2 months. Adoption took 4. If you budget an FDE only for the build, you'll get a system that works and nobody uses.

FDE vs solutions engineer, consultant and software engineer

The titles overlap, and plenty of companies post "FDE" for a different job. Bloomberry read 1,000 FDE job postings and found the title used for 3 different jobs, and 0% of them carried a sales quota (Bloomberry). Use this table to check what you're actually hiring.
Role
When they work
What they deliver
Writes production code?
Solutions / sales engineer
Before the sale
Demos, proofs of concept, answers to technical questions
Rarely
Consultant
Project by project
Recommendations the customer then builds
Rarely
Software engineer
Ongoing, on the core product
Features for all customers
Yes, on the product
Forward deployed engineer
After the sale, embedded
A working system in the customer's environment, plus product feedback
Yes, on the customer's systems
Short version: a solutions engineer sells it, a consultant explains it, an FDE ships it.

Why forward deployed engineers are in demand in 2026

Models got good, deployment didn't

Most companies already have access to strong models. Few get value from them. MIT's NANDA study of 300+ AI initiatives found 95% of organizations saw no measurable return from generative AI pilots (The Journal). The sample was small and success was narrowly defined, but the cause it names matches what FDEs see: the tools didn't fit the workflows and didn't learn from them. That's integration work, not model work.
"People were really excited, but it was also, like, kind of hard to get value from the models," said Colin Jarvis, who runs OpenAI's FDE team (Business Insider).

Agents have to plug into messy systems

A chatbot answers questions. An agent files the claim, updates the CRM and emails the customer. That means access to 5 internal systems, each with its own permissions, odd data and owner. Somebody has to sit inside the company to wire it up, and that somebody has to be an engineer.

The AI labs and clouds are building FDE teams

  • OpenAI had 39 FDEs in late 2025, planned to reach 52 by year end, and posted US salaries up to $345,000 plus equity. Its projects are worth "tens of millions to sometimes the low billions" (Business Insider).
  • Anthropic hires FDEs to embed with its largest customers and build MCP servers, sub-agents and agent skills for production workflows (Anthropic posting).
  • AWS announced a $1 billion forward deployed engineering organization on June 30, 2026, to place engineers inside customer teams and build agentic systems with them (The New Stack).
  • Palantir, Ramp, Salesforce, Commure and hundreds of AI startups hire for the role too. Ramp runs about 15 FDEs in pods (The Pragmatic Engineer).

Services first, product later

Andreessen Horowitz called the FDE "the hottest job in startups" in June 2025 (a16z). The argument: enterprises buying AI need someone to set it up, so AI startups accept lower margins early to get deeply embedded. ServiceNow went public at a 63.2% gross margin and Workday at 54.1%; by 2024 they were at 79% and 75%. The setup work became the moat.
As a16z puts it: "Enterprises buying AI are like your grandma getting an iPhone: they want to use it, but they need you to set it up."

AI writes the code, so the scarce skill is knowing what to build

Claude Code, Cursor and Codex now write much of the code, so the slow part is no longer typing. It's understanding the business: which process to automate, which data to trust, what "done" means for the ops manager. That is exactly the FDE's job.

Do you need a forward deployed engineer?

2 kinds of companies hire FDEs.
  1. AI startups selling to larger companies. Your product works in the demo but each customer has different systems. FDEs get each customer live, then turn what they built into product.
  1. Companies putting AI into their own operations. You don't need another tool. You need 1 engineer who sits with your team, finds the 3 workflows worth automating and ships them.

What a forward deployed engineer costs

Source
What it shows
Bloomberry, 1,000 US postings
$173,816 median base salary; 70% of postings mention equity
US salaries up to $345,000 plus equity
Mostly $200,000-300,000 base in the US
Paralect Match, contract
$30-60/hour by seniority, about $4,800-9,600 for a full-time month
A full-time US FDE at the median costs about $14,500 a month in base pay alone, before equity, benefits, recruiting fees and the 2-3 months it takes to hire. For an AI lab with $10 million deals, that's cheap. For a startup with its first 5 enterprise customers, or a company with 3 workflows to automate, it's often the reason the project never starts.
Pay figures checked in November 2026. Salaries for this role move fast; confirm on the postings before you make an offer.

Skills to look for

From Bloomberry's 1,000 postings: Python appears in 66%, TypeScript in 35%, AWS in 32%, AI agents in 35% and LLMs in 31%. The 3 most common duties, each in 60%+ of postings: working directly with customers, building and deploying AI systems, and integrating systems and APIs.
The code is the easy half to check. These are the skills that decide whether the project ships:
a16z's advice fits: hire "curious hustlers." The job "doesn't require a PhD, but it does require hustle."

How to vet a forward deployed engineer in 1 week

A CV can't show whether someone can walk into a messy business and ship. A short paid trial on a real problem can.
Day
Do this
You learn
1
Write a 1-page brief: the workflow, the systems, the result you want
Whether you know what you're asking for
1-2
Ask for 2 projects they deployed for a customer or team, with what changed after
Who has shipped beyond demos
2-3
45-minute call with the 5 questions below, with someone from the team they'll work with
How they talk to non-engineers
3-5
Paid 2-day trial on a slice of the real problem, with real (or realistic) data
Whether they ask the right questions and ship
5
Review: did it run, is there an eval, did they explain it in plain words?
Your decision

5 interview questions for an FDE

1. "Tell me about a deployment that almost failed. What happened?"
  • Good: a specific story about data, access or people, and what they changed.
  • Red flag: only technical problems, or "nothing failed."
2. "How did you measure that the AI part worked?"
  • Good: a set of real cases, a score, a threshold agreed with the customer before launch.
  • Red flag: "Users liked it." "We tried a few prompts."
3. "Explain your last project to our ops lead." (Have them in the room.)
  • Good: plain words, starts from the ops lead's problem, asks them questions back.
  • Red flag: jargon, architecture diagrams, no questions.
4. "What did you build for one customer that ended up in the product?"
  • Good: a connector, template or feature other customers now use.
  • Red flag: every project was one-off. That's a contractor, not an FDE.
5. "What did you say no to?"
  • Good: a request they turned down or postponed, and why it would have slowed the result.
  • Red flag: they built everything they were asked for.

Red flags

Write the brief before you hire

Most FDE hires go wrong before the first interview, because the company can't say what it wants. Fill this in first; if you can't, you're not ready to hire anyone.

Loading editor...

Outcome:     What changes when this works? (e.g. "support answers refund requests in 2 minutes, not 2 days")
Workflow:    Who does this today, step by step?
Systems:     Which tools and data does it touch? (CRM, ERP, Google Drive, Postgres...)
Access:      What can the engineer see on day 1? Who grants the rest?
People:      Who will the engineer sit with? Who signs off?
Measure:     How will you know it works? 20 real cases with the right answers is a good start.
First task:  The 2-day slice you'd give as a trial.

Hire a forward deployed engineer through Paralect Match

Paralect Match gives you an AI-native engineer who works the way an FDE does: inside your team, with your people, accountable for what ships. It's a fit for both cases above: getting your customers live, or putting AI into your own operations.
  1. Send a short brief. The one above is enough.
  1. Get 2-3 engineers in 48 hours, each with their hourly rate and a demo of something they shipped. Every engineer is interviewed by Paralect's CTO and has built and presented a complete full-stack MVP to get in.
  1. Run a 2-day trial on a real task for $7. You see how they ask questions and what they ship before you commit.
  1. Continue by the hour, $30-60 depending on seniority. The engineer joins your Slack and sends daily updates. Weekly billing, no long-term contract, cancel any time.
Engineers on Match work remotely. If the job needs someone at a customer's office every week, plan for travel or hire locally.
Find your forward deployed engineer on Paralect Match: 2-3 vetted candidates in 48 hours, a 2-day trial for $7.

FAQ

What is a forward deployed engineer?
A software engineer who works embedded with one customer or team, writes production code on their systems, and stays until the system is live and used. The role started at Palantir in the early 2010s, where it was called "Delta."
What does a forward deployed engineer do day to day?
Talks to the people who'll use the system, maps their workflow and data, writes integrations, agents and evals in the customer's environment, helps people adopt it, and feeds what they learned back to the product team. Palantir expects about 25% of the time onsite.
How much does a forward deployed engineer make?
The median base salary in US postings is $173,816, and 70% of postings include equity (Bloomberry). OpenAI's US postings go up to $345,000 plus equity. On a contract through Paralect Match, an engineer costs $30-60 an hour.
What's the difference between a forward deployed engineer and a solutions engineer?
A solutions engineer works before the sale: demos and proofs of concept. A forward deployed engineer works after it, writing production code on the customer's systems until it runs.
Why are forward deployed engineers in demand?
Companies have access to strong AI models but struggle to get them working in real workflows. Agents need deep integration with internal systems, and that takes an engineer inside the business. OpenAI, Anthropic and AWS have all built FDE teams, and monthly FDE listings grew 800%+ from January to September 2025.
How do I hire a forward deployed engineer quickly?
Write a 1-page brief, ask for 2 deployed projects, run a call with someone from your team, then a 2-day paid trial on a real problem. Through Paralect Match, you see 2-3 candidates in 48 hours and run the trial for $7.

Sources

Ship faster than
your competition

Focus on customers and sales while we handle product delivery. Hire a dedicated AI maker or a whole product team.

Launch new products
Fix your delivery
Hit fundraising milestones
[Analytics Chart]
Wireframe v0.1
User
Sarah Chen CEO, Founder
$127k
MRR
2,847
Customers
+23%
Growth
Sales Performance
Latest Deal
Customer
Enterprise plan closed
Acme Corp · $12k/mo
© 2026 Paralect, Inc 651 N Broad St, Suite 206, Middletown, 19709, Delaware, United States