AI Engineer
Designed and deployed a production RAG pipeline for personalised entertainment recommendations across millions of items, running on Cloudflare’s edge stack — Workers, D1, KV, Vectorize, Cron Triggers — with a Next.js front end. Built agentic workflows with multi-provider LLM integration: autonomous semantic routing, query decomposition, and dynamic tool calling. Put the whole thing on Langfuse for prompt-trace tracking, retrieval-quality monitoring, and regression checks on every prompt or retrieval change.