You already know: prompting, RAG and agents (D7–D9); evaluation (D10); cloud and security basics (D17, D19); decision-making and writing (this domain).
Why this matters for Atlas. AI products rarely fail in the demo. They fail in the gap between a demo and a specific customer's reality: their data, their systems, their rules, their people. The forward deployed engineer (FDE) is the role built to close that gap. It's one of the fastest-growing roles in AI engineering because the gap is where most AI value is won or lost. This lesson explains what the job really is, the loop you run every week, the skills that matter, the traps that sink good engineers, and how the role is measured. The next lesson turns it into practice: scoping a deployment with a customer.
1. The job in one sentence
The question. Is an FDE a consultant, a solutions engineer, or a software engineer who travels?
Start from nothing. An FDE owns one customer's outcome, end to end, and turns what they learn into product. Both halves matter: