cv-santiago

santifer.io is an interactive portfolio and CV that demonstrates production AI engineering skills through a working chatbot named "Santi." The project combines a React 19 and TypeScript frontend with Vercel Edge functions, Claude Sonnet for text chat, and OpenAI Realtime API for voice mode. The chatbot uses agentic retrieval-augmented generation with hybrid search (pgvector plus BM25) and Claude Haiku reranking. The repository includes a private LLMOps dashboard at /ops with eight tabs covering overview, conversations, costs, RAG, security, evals, voice, and system data sourced from Langfuse and Supabase. It also provides 71 automated evaluations across ten categories, a six-layer prompt-injection defense with jailbreak email alerts, and a closed loop that generates tests from low-quality production traces. Additional features include bilingual English/Spanish case studies with JSON-LD and prerendered HTML, an interactive GSAP-animated architecture diagram, prompt versioning through Langfuse, and numerous CLI scripts for chatbot operations, prompt and RAG management, contract testing, and build validation. The project is MIT licensed and designed to run on free infrastructure tiers.