Christopher Alatorre
Principal Frontend Engineer
Fifteen years building production frontends. The work has taken me through high-concurrency commerce, AI tooling adoption, and embedding-based retrieval systems. Now I'm building a RAG pipeline from scratch.
The Work That Got Me Here
Most recently I was a Principal Web Engineer on the SNKRS team at Nike. SNKRS runs high-concurrency launch events. High demand Low inventory in a small time window, and the system has to hold. My work lived in that pressure: reliability at burst traffic, checkout consolidation across consumer surfaces, keeping the experience coherent when the load is anything but.
I led AI tooling adoption workshops for engineering teams — working through backlogs using spec-driven frameworks like spec-kit, running skill-writing sessions, and fielding questions in Slack channels dedicated to the rollout. Helping engineers figure out where these tools actually fit versus where they add noise.
I also built rules-lawyer — a local rules retrieval project using embedding models. That work gave me direct exposure to what retrieval quality means when the domain is dense and structured. Greenskin Labs is the production version of that question: proper ingestion pipeline, hybrid retrieval, citation grounding, built to run in the cloud.
What I'm Building Now
Greenskin Labs is a set of tools for tabletop wargamers. The Rules Query engine is a RAG pipeline over a wargame ruleset — layered books, errata, supplements, the kind of corpus that makes a good search surface genuinely useful.
The architecture: Google Document AI for PDF ingestion, pgvector in Supabase for dense retrieval, BM25 for sparse, a hybrid reranker to combine them. Answers are grounded in cited chunks with supersession detection so updated errata don't surface stale rulings. LLM calls proxy through Cloudflare AI Gateway for rate control and observability.
The blog documents the decisions — why hybrid over pure dense, how supersession detection works without a structured changelog, where the pipeline breaks. This is a tool I use.
Resume and references available on request. greenskinlabs@proton.me