Forward Deployed Engineer (FDE) in the AI Ecosystem

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

The Forward Deployed Engineer (FDE) role has proliferated across artificial intelligence (AI) companies since 2023, yet no formal engineering-role definition, competency framework, or comparative analysis distinguishes it from four adjacent roles it is routinely confused with: Software Engineer (SWE), Machine Learning (ML) Research Engineer, Solutions Architect, and Sales/Systems Engineer. This absence produces measurable organizational cost: job descriptions copied from Solutions Architect templates, compensation benchmarked against sales scales never designed for production-code-writing engineers, and hiring bars that underweight the discovery and integration skills the role actually requires. This paper proposes a six-dimension taxonomy — primary output, client interaction level, ambiguity tolerance, deployment-versus-research focus, travel expectation, and reporting line — that formally separates FDE from its four neighbors, grounded in current hiring evidence. Building on this taxonomy, we construct a technical, client-facing, and organizational competency framework, and a four-stage engagement lifecycle model (discovery, prototype, integration, handoff) with an explicit feedback loop connecting integration findings back to discovery. We then compare how Palantir, OpenAI, Anthropic, and Scale AI structure and deploy forward-deployed teams, using only public source material, and find genuine structural convergence around embedded, product-modification authority alongside a documentation asymmetry across the four companies. We name five limitations explicitly, foremost the absence of primary practitioner survey data, and close with a research agenda that treats empirical validation of the framework as the field’s highest-priority next step.