Most engineers who want the Forward Deployed role already have the raw material. What they're missing isn't more knowledge — it's proof that survived contact. Here's how to build it in a quarter.
Weeks 1–2 — find your two weakest pillars. The job rests on six: production instinct, full-stack range, LLM/agent fluency, client nerve, deployment temperament, proof of work. You're probably strong on three and thin on two. Name them honestly (the diagnostic does this in three minutes), and spend the quarter on the two, not the six.
Weeks 3–8 — ship one real deployed system. Not a demo. One system with all the FDE ingredients: an LLM or agent doing real work, wired into a messy external API with auth and rate limits, in front of real users, with an eval set that shows it works. Keep it alive for 30 days. The mess you fight is the curriculum.
Weeks 9–11 — document and position. Write it up: the architecture, the legacy you fought, what "good" meant and how you measured it. This artifact is your candidacy. Turn it into a narrative — "I deploy AI in production, in the real world" — and lead your portfolio with shipped systems, not certificates.
Week 12 — the interview. FDE interviews test the intersection: system design under real constraints, a client-facing role-play, eval-scoping for a fuzzy task, and three deployment war stories. Have them ready.
Ninety days won't make you senior. It will make you employable — someone with a real deployment to point at, which is the only thing the role actually screens for.