How to Become a Forward Deployed Engineer

The forward deployed engineer role is attractive for good reasons: it pays well, it is varied, and you see the effect of your work directly. It is also easy to misread as "a software engineer who talks to customers". The customer part is the job, not a side duty.

I work as a freelance FDE, after four years of AI and full-stack client work across the US, UK, Europe, the Middle East, Australia and India. This guide is the path I would recommend to an engineer starting now, with a six-month plan and the portfolio projects that prove each skill. If you need the definition first, start with what a forward deployed engineer is.

Where forward deployed engineers come from

Most forward deployed engineers move into the role from software engineering, not straight from university. Bloomberry looked at 100 LinkedIn profiles of current FDEs and found their previous roles were software engineer (45%), solutions engineer or architect (22%), data engineer or scientist (15%), technical consultant (10%) and founder or early employee (8%).

The same analysis of about 1,000 job postings found 60% ask for three to five years of experience, 20% for six to eight, and only 12% for two years or less. There are exceptions. Gergely Orosz reports that Palantir hires FDEs with as little as one year of post-college experience, while Ramp prefers five or more years for senior roles.

So the realistic routes are:

• From software engineering: the most common. You already ship code; you need customer exposure and AI system experience.
• From solutions or sales engineering: you have the customer side; you need to prove you can own production code, not just demos.
• From data roles: you have the data skills; you need full-stack range and integration work.
• From founding or freelancing: you have done everything badly at least once, which is surprisingly good preparation.

The four skills: code, customers, AI and judgment

Forward deployed engineering needs four skills at once, and hiring managers test all four. Missing one is usually what ends a candidacy.

Job postings back this mix. Bloomberry found the most common responsibilities are working directly with customers (55%), building and deploying AI or ML systems (37%) and integrating systems and APIs (32%). A since-closed Anthropic FDE posting asked for production experience with LLMs, including prompt engineering, agent development and evaluation frameworks, alongside strong Python and "strong communication skills for customer discovery".

The customer and judgment skills are the hardest to show on a CV. The recruiting firm Paraform tells hiring managers that customer empathy and ambiguity tolerance are the hardest to screen for, and that a green flag is a candidate who "frames past work by customer outcome, not implementation". Practise describing your work that way now.

A six-month plan to become a forward deployed engineer

If you already work as a software engineer, six months of focused evenings and weekends is enough to build a credible FDE portfolio. The plan below adds one skill at a time and ends with a real client project.

Month 1, production fundamentals. Ship one full-stack application to production: authentication, a database, a background job, logging and a deployment you can repeat. If you already do this at work, skip ahead.

Month 2, the AI building blocks. Build retrieval over a real document set, and get structured output from a model validated against a schema before you save it. Learn where it fails by reading the bad outputs, not the good ones.

Month 3, integrations and evaluation. Give the system tools: let it read from and write to a real system through an API, or build an MCP server for one. Write an evaluation set of real examples and run it on every change.

Month 4, a real user. Find someone with a real problem: a small business, a non-profit, a team at a friend's company. Run a short discovery: interview them, watch them work, and agree what success means. My FDE discovery playbook describes the steps. Then build the smallest thing that solves it.

Month 5, production hardening. Add what turns a demo into a system: error handling, logging, a review step for risky actions, a monthly cost estimate, and a handover document so they can run it without you.

Month 6, evidence and interviews. Write up each project as a short case study: the problem, what you built, what changed, what broke. Practise decomposition questions out loud. Then apply.

Portfolio projects that prove FDE ability

The best FDE portfolio projects are small systems someone actually used, each one proving a different skill. Three good ones beat ten tutorials.

• A document workflow for a real team. Extraction from messy inputs, validation, a review screen, and a measured result. Proves code, AI and customer work together.
• An integration that writes back. An agent or service that reads from one system and updates another with permissions and an audit log. Proves you can be trusted inside someone else's stack.
• An evaluation harness. A test set of real examples and a script that scores the system on every change, with a chart of results over time. Proves you know how to tell whether AI is working.
• A written scope and handover. The discovery notes, scope and handover document from your client project, with anything sensitive removed. Proves judgment, which almost nobody shows.

When you describe these, lead with the user and the outcome, then the stack. That is the format interviewers listen for.

Getting hired: titles, routes and expectations

Look beyond the exact title. Many companies hire for the same work as AI engineer, solutions engineer, applied AI engineer or customer engineer, and some roles titled FDE are closer to sales support. Bloomberry classified about 30% of the postings it read as sales-engineer roles with an FDE title, so read what the job actually asks you to build.

Expect travel and expect variety. In Bloomberry's sample, 68% of postings mention travel, and 58% of FDE roles were at companies with 11 to 200 employees, which means small teams and broad responsibility.

Freelancing is another route in, and the one I work in today. It gives you customer exposure fast, but you carry the sales and the risk yourself. If you go that way, the career ideas in how to become an AI engineer and the pay data in forward deployed engineer salary and demand are useful companions.

Work with me

If you are a company that needs forward deployed engineering now rather than in six months, I take on fixed pilots and longer engagements. See AI agent development and full-stack development, look at my work, and book a scoping call.