Good news: you don't have to move to San Francisco.
Anthropic, OpenAI and Mistral are all hiring Forward Deployed Engineers (FDEs) in London and Paris, and Anthropic and OpenAI in Munich too. OpenAI also owns a UK consultancy with about 150 of them. Even the UK Ministry of Justice has a team.
We read the job ads, the hiring pages and the interview stories people share online, and mapped it all out.1
In this article you'll find answers to:
- Who hires FDEs in Europe, and where
- What the interviews look like, and what to check before you apply
- Whether you still need LeetCode
Who hires
We found more than 30 companies with FDE roles in Europe. They fall into six groups.
| Kind of employer | Examples |
|---|---|
| AI labs | Anthropic, OpenAI, Mistral, Cohere |
| Big Tech and data platforms | Google Cloud, Microsoft, Palantir, Databricks, Snowflake |
| AI scale-ups | ElevenLabs, Parloa, Synthesia, Intercom, Lovable, n8n, Langdock, Zauber |
| Defence | Palantir UK Government, Helsing, ICEYE |
| Public sector | UK Ministry of Justice, Intric (a Swedish AI platform used by municipalities) |
| Consultancies | NTT DATA, BCG X, Sii, Tomoro (now part of OpenAI) |
London has the most by far. Labs, Big Tech, defence, consultancies and the only government FDE team we found all hire there.
Berlin is mostly European scale-ups: n8n, Langdock, Parloa, JetBrains, Zauber. Paris has the labs plus French startups like H Company and HappyRobot. Munich has FDE roles at Anthropic and OpenAI. Amsterdam, Zurich, Madrid, Warsaw, Stockholm and Dublin each have a few.
Speaking the local language helps, and sometimes it's required. When we looked in October 2026, Anthropic's Paris and Munich roles asked for native French or German, Langdock wanted German at C1, and Helsing wanted German for roles in Germany.
Many roles are senior, too. Anthropic called its European FDEs "founding" hires and asked for 8+ years in Paris and Munich. Intercom and HappyRobot also advertise founding roles.
Interviewing from Europe
Interviewing from London or Munich looks much like interviewing from San Francisco. No lab publishes a separate process for Europe. OpenAI's interview guide describes one process for everyone: an intro call, a skills assessment ("pair coding interviews, take-home projects, technical tests"), then "4–6 hours of final interviews with 4–6 people over 1–2 days." It's virtual by default, so you don't need to fly anywhere.
Three things are different, though.
- Language. Some roles need a local language for customer work even when the ad is in English.
- Location and visas. Some roles listed in Europe end up being hired somewhere else. Anthropic's London ad said it sponsors visas but can't "for every role and every candidate."
- Team matching. At big companies, passing the interviews doesn't always mean an offer. You may still need to find a team, and the team in your preferred city can fill up while you wait.
Two kinds of interview
Most FDE interviews are one of two kinds.
Build something, then defend it. You get a short build task or a take-home, often with an LLM you're allowed to use. Then a panel or the team asks why you built it that way, how you'd extend it and what it would take to run in production. If you leaned on AI and don't understand the result, that shows quickly. Take-homes run from a few hours to a week, and they look a lot like the job.
Zauber, a Berlin startup building AI agents for freight forwarders, publishes its interview process for Product Engineers. The role is close to FDE, since you also talk directly to customers. Their process: a 15-minute call, a 15–30-minute deep dive on something you built, 30–60 minutes of system design, then a half-day on-site. There you build for five hours "using whatever tools you prefer," demo it to the team, do a product interview and have a team lunch or dinner. It usually takes about a week. Their advice: "Be AI-pilled."
The classic coding screen. Big Tech and enterprise software companies, like Google and Salesforce, still include an algorithms round or an online coding test, alongside a business case and a design round.
AI rules differ between companies, and sometimes between rounds. Anthropic's policy is "no AI assistance unless we indicate otherwise." Others hand you an API key. Ask before each round.
Is LeetCode dead?
Not dead. It's moved from the main test to a filter.
In most FDE interviews, LeetCode-style questions show up early and are on the easy side, sometimes with AI allowed. Data platforms may test applied data work instead, like pandas and classical machine learning.
The rounds that decide whether you get the job come later: the build, the defence and the customer case.
Google and Salesforce are the main exceptions we found. Their FDE interviews still include a real algorithms round, so prepare as you would for a software engineer interview.
Questions to ask the recruiter
- Can I use AI tools in each round? Which ones? Rules can change from round to round, even inside one company.
- Is there a take-home? How long should it take? Good sign: a clear time limit.
- Which language will I work with customers in? Ask even if the ad doesn't mention one.
- Can this role be hired in my city? Do you sponsor visas? Ask early, not after the final round.
- After I pass, is there team matching? If there is, ask how long it usually takes and which cities have open teams.
- Is the FDE role on the same ladder and band as software engineers? Is the pay all base salary, or partly tied to targets? Either can work, but it changes how much of your pay is guaranteed.
So, who hires FDEs in Europe?
More than 30 companies, mostly in London, then Berlin and Paris.
Before you apply, find out whether you'll build something and defend it or sit a classic coding screen. Then sort out the European parts early: the language, the visa and the team matching.
Notes
1. Very little of this is published by the companies themselves. Most of what's out there is people describing their own interviews, and few of those are from Europe. We read them for patterns and didn't lean on any single one.