How to Become a Forward Deployed Engineer: 2027 Career Roadmap
A practical roadmap to becoming a forward deployed engineer, covering the skills, salary range, hiring companies, and steps to land the role.

What is a forward deployed engineer?
A forward deployed engineer (FDE) is a software engineer who works directly with a customer, on site or close to it, to figure out what that customer actually needs from an AI product and then build it into their existing systems. The role started at Palantir and has since spread to nearly every major AI company. Instead of selling a login and hoping the customer figures it out, the company sends an engineer to sit with the customer’s team, scope the real problem, and ship something that works in production. As one former Palantir FDE now at Anthropic put it, the role is a customer-facing software engineer, and the customer isn’t buying software or hours, they’re buying an outcome.
TL;DR
- Forward deployed engineer roles pay a typical base of roughly $180,000 to $275,000, with median total compensation (including stock and bonus) landing around $230,000 once you factor in postings from 24 AI companies.
- The job exists because most companies can’t operationalize the AI tools they buy, a gap that lines up with the widely cited finding that a large majority of enterprise generative AI projects produce no measurable return.
- Discovery and stakeholder communication show up in roughly 80% of FDE job postings, which means the role is less about raw coding ability and more about diagnosing real problems and managing people through change.
- Python appears in about 60% of postings, but coding is treated as a baseline skill rather than the differentiator, especially now that tools like Claude Code and Codex can generate most of the implementation.
- Evals are the most underrated skill in the stack: they’re mentioned in about half of postings, but they’re the only way to actually prove to a customer that an agent or automation works, rather than just demoing well once.
- Most people break into this role laterally, from software engineering, consulting, solutions engineering, or running their own AI agency or freelance practice, rather than through a dedicated junior FDE track.
- The title might not survive the decade the way “prompt engineer” didn’t, but the underlying skill loop (scope a problem, build it, prove it, make it repeatable) is durable even if the job name changes.

What you'll learn · Dec 2
- 01How enterprises really build with AI
- 02What full-stack means and why it's critical for business software
- 03Prompting that holds up in production
- 04Build and deploy your own app, live
Why is every AI company hiring forward deployed engineers right now?
Companies are buying AI software faster than they can actually implement it. That’s the gap FDEs are hired to close. The MIT report that found 95% of companies saw zero return from their generative AI spending captures the problem: enterprises write checks for AI platforms, then stall out because their data is messy, their workflows don’t match the tool, and nobody internally knows how to bridge the two.
AI companies responded by building out forward deployed engineering teams that embed with customers post-sale. Anthropic’s own FDE postings describe deliverables like MCP servers, sub-agents, and agent skills, essentially the same building blocks a developer might use inside Claude Code, just applied to a paying enterprise customer with real stakes and real deadlines. OpenAI, Palantir, Scale AI, Cursor, Anthropic, Cohere, Cognition, Databricks, and Google Cloud all have active FDE hiring, and job postings for the role have grown more than 10x year over year.
What does the job actually involve day to day?
Based on job postings and firsthand accounts, a typical week looks something like this: Monday is spent on site doing discovery, sitting with the people who actually do the work and asking questions. Tuesday is scoping, writing down the problem, the metric you’re trying to move, and what “done” means. Wednesday and Thursday are build days, wiring an agent into the customer’s CRM or data systems and writing evals to test it against real data. Friday is demo day, presenting to stakeholders and feeding notes back to the product team about where the model broke or what customers keep asking for.
Colin Jarvis, who leads forward deployed engineering at OpenAI, has said new hires should expect to spend around half their time on site with customers. Anthropic’s postings also ask FDEs to “identify and codify repeatable deployment patterns,” meaning part of the job is turning a one-off solution into something reusable across other customers, not just solving the same problem once and walking away.
What skills do you actually need?
Despite the word “engineer” in the title, the job is a hybrid. Across roughly 200 live postings pulled from 24 AI companies, discovery or scoping work appeared in about 80% of listings, stakeholder and executive communication in about 80%, and customer-facing work in roughly three-quarters. On the technical side, Python showed up in about 60% of postings, TypeScript or JavaScript in about 60%, and APIs and integrations in about 60%, since the work almost always involves connecting one system to another. “Agents” appeared in around 60% of postings, referring to systems where a model decides what to do next and calls tools across multiple steps, which is a step beyond just using an AI coding assistant to write code.
Evals showed up in only about half of postings, but they matter more than that number suggests. A demo proves a system can work once. An eval proves it works reliably, which is what a customer and their stakeholders actually need before they’ll trust an automation in production. Anyone skipping evals and just shipping a working demo is doing the AI equivalent of vibe coding, and it tends to fall apart once real usage starts.
Most postings ask for five or more years of experience, only about one in ten accept three years or less, and only about one in five even mention a degree requirement. Anthropic explicitly says former technical founders are encouraged to apply, which signals this is treated as a lateral move rather than an entry-level track.
Who is actually a good fit for this role?
People land FDE roles from roughly three backgrounds: software engineers who genuinely like talking to people and sitting in messy meetings, consultants or solutions engineers or analysts who’ve picked up enough technical skill to build, and technical founders, freelancers, or agency owners who’ve effectively been doing this work without the title. If you’re already running an AI automation agency or consultancy for small businesses, you’re functionally doing FDE work already, just for a different kind of customer.
It’s not for everyone. About two-thirds of postings mention travel, usually in the 25% to 50% range. One recruiter quoted in Pragmatic Engineer noted that everybody wants to hire FDEs, but only around 10% of the market actually wants to do the job. If you like ambiguity, you like sitting with a customer’s operations team and asking a lot of questions, and a messy, half-defined problem doesn’t scare you off, that 10% is where you want to be.
How do you actually become a forward deployed engineer?
Start by being honest about which half of the role you already have: the builder half or the trusted-advisor half. Engineers usually need to work on discovery and executive communication. People coming from consulting, operations, or analytics usually need to work on building things that hold up in production. Spend most of your prep time on your weaker half instead of studying everything evenly.
A cheap way to practice discovery is to have an AI model roleplay a difficult customer. For example, prompt it to act as a VP of operations at a mid-sized company that bought an AI platform three months ago and has nothing working, and have it refuse to volunteer information until you ask good questions. After the session, ask it to grade your discovery and point out what you missed. Follow every practice session by writing a one-page scope: the problem in the customer’s own words, the one number you’re trying to move, what “done” looks like, and what’s explicitly out of scope. If you can’t name the number you’re moving, you don’t have a real project yet.
If you’re currently employed, volunteer for demos, onboarding calls, or requirements meetings at your own company. If not, pick one project, your own team, a friend’s business, a local company, and run the entire loop: discovery, scope, build, evals, demo, then follow up two weeks later to measure whether people are actually using what you built. Write that up as a short before-and-after case study. That case study becomes the thing you walk interviewers through.
Where are FDE jobs actually posted?
Built like a system. Not vibe-coded.
Remy manages the project — every layer architected, not stitched together at the last second.
Most companies post these roles on standard applicant tracking systems like Greenhouse, Ashby, and Lever, or directly on their own career pages. Based on current postings, the companies with the most open FDE roles include Databricks, Palantir, OpenAI, Scale AI, Cursor, Anthropic, Cohere, and Cognition, along with a number of smaller AI startups. Google Cloud also hires for the role through its own career site. VC job boards from firms like Andreessen Horowitz, Accel, and General Catalyst are useful too, since they aggregate openings across their portfolio companies in one place. It also helps to search beyond the exact title “forward deployed engineer,” since the same job shows up as forward deployed software engineer, deployment engineer, or deployment strategist for less technical versions of the role.
Frequently Asked Questions
What is the salary range for a forward deployed engineer?
Based on roughly 200 live postings across 24 AI companies, typical base salary runs about $180,000 to $275,000, with median total compensation (including stock and bonus) around $230,000.
Do you need a computer science degree to become an FDE?
No. Only about one in five postings mention a degree requirement at all. Companies care more about whether you can run the full loop of scoping a problem, building a solution, and proving it works.
Is forward deployed engineering a coding-heavy job?
Coding is a baseline requirement, not the differentiator. Python appears in about 60% of postings, but AI coding assistants handle much of the implementation now. The harder-to-find skill is discovery and stakeholder communication, which shows up in about 80% of postings.
Will the forward deployed engineer title still exist by 2027?
That’s uncertain. Titles like “prompt engineer” became hot and then faded once the underlying skill got commoditized. FDE work is harder to compress into a single skill because it’s a full loop of discovery, building, and proof, but the job could still end up under a different name, such as AI consultant or deployment strategist.
How much travel does the job involve?
About two-thirds of postings mention travel, typically in the 25% to 50% range, and some FDEs report spending around half their time on site with customers.

