How to become a forward deployed engineer

How to become a forward deployed engineer: the short answer

Forward deployed engineer interviews test practical coding, system design inside a customer's environment, breaking down open-ended problems and stories about work you owned. Certificates are not on that list.

This page sets out the stages in order. Each one has the skills, one project to build and good places to learn.

Companies can also hire forward deployed engineers through us. They tell us what they need, and we send people who fit.

Skills employers ask for

Anthropic's Applied AI post is a useful benchmark because it spells out the bar. It asked for years in a technical, customer-facing role, production experience with LLMs, strong Python and strong communication. It also listed travel to customer sites.

Palantir describes its forward deployed engineers as the people who make many capabilities work for one customer. Read each stage with that in mind. Breadth, speed and ownership matter more than depth in one narrow area.

Foundations

Goal: write correct code fast in one language, and be at home in a terminal.

Skills: Python to a professional level, including testing. Git beyond add and commit. The shell, ssh and reading logs on a remote machine. SQL basics. Enough algorithms to pass a standard coding screen.

Build this: a command-line tool that reads a messy CSV export from a public dataset, cleans it with clear rules, loads it into SQLite and prints a report. Include tests and a file of the rows that failed.

Shipping full-stack apps

Goal: put a working app in front of real users and keep it running.

Skills: a backend web framework and a basic frontend. Postgres schema design, migrations and indexes. Sign-in and the basics of OAuth. Docker and one cloud provider. Logging and monitoring, so you know it broke before the user tells you.

Build this: a small internal tool for a real person you know, such as a booking app for a club. Deploy it, give them logins and support it for a month. Write down every bug they hit.

Data and integrations

Goal: connect systems you did not build, and move data between them without losing any.

Skills: calling third-party APIs, with paging, rate limits, retries and OAuth. Webhooks and idempotency. SQL at working depth, including window functions and slow query fixes. Data pipelines with quality checks. Reading CRMs, ERPs and data warehouses.

Build this: pull data from public APIs that use different sign-in and paging styles. Land it raw, clean it into tables with dbt or plain SQL, and add checks that fail loudly when a source changes shape.

LLM apps in production

Goal: build AI features that hold up with real users, and prove it with evidence.

Skills: prompting and structured output with a major model API. Retrieval, including chunking, embeddings, hybrid search and access control. Agents and tool use, with narrow tools and human approval for writes. Evals built from real inputs. Cost and speed: caching, model routing and trimming context.

Build this: a question-answering assistant over a real document set, such as council minutes or an open-source project's docs. Ship it with citations, an eval set of real questions and a short report on what it gets wrong.

Customer-facing skills

Goal: find the real problem, write it down and keep a customer's trust when things slip.

Skills: discovery interviews that end with a workflow, an owner and a measure of success. Writing a short proposal, a design doc and a weekly update. Cutting a big ask down to a first version. Giving bad news early, with options. Handling incidents without blame.

Build this: find a small business or charity with a manual process. Interview the people who do the work, write a short proposal with a measure of success, build the first version and send written updates until it is in use.

A portfolio of deployments

Goal: show proof that you took something from a vague ask to production use.

Skills: write-ups that lead with the problem, your role and a measured result. Demos and repositories a reviewer can run quickly. Short ownership stories for interviews, told with "I".

Build this: turn your projects into short case studies. Cover the user, the problem, what you built, what changed and what you would do differently. Link a demo or repository for each, and send it with every application.

Forward deployed engineer courses and certificates

Search for a forward deployed engineer certification and you will find several paid programmes. Some may give you useful structure. But none of the job posts we read asks for a certificate, and the role is too new for any credential to be a standard.

What hiring managers look for instead: systems you shipped that real people used, time spent with customers, LLM work beyond a demo, clear writing and a clean practical interview.

Getting hired once you are ready

Check what the role pays so you know what to ask for. Then apply where your portfolio fits. Our jobs page links to roles on companies' own careers sites.

You can also apply to us once with your CV, for fractional and contract work. An automated screen helps us sort applications, then we look at each one ourselves. Talent never pays.