Infrastructure and operations engineer with 20+ years of experience, now specializing in AI/GPU infrastructure operations across managed-services programs…
Ask the résumé anything. Answers are grounded in the CV by a small, private, self-hosted language model that streams token-by-token — with deterministic retrieval as an always-on fallback, so it never hard-fails. The engineering mirrors the work it describes: local models on owned hardware, data residency by default.
Full-text search, five export formats, career stats and a unified timeline — every route documented in an interactive OpenAPI spec at /docs. The 2020 v1 contract still answers at /app/api/v1.0/cv.
Public v2 reads plus subscriber/admin token tiers — JWT, API keys and the legacy bearer, all wired. Rate-limited, CORS-clean, monitored like everything else he runs.
Docker + Traefik + automatic Let's Encrypt TLS. The language model runs in its own isolated, memory-capped side-container — cv-llm — on the same Docker network.