Case Study: Emerald City Beacon
This wasn’t just a chatbot glue job, it’s a full-stack RAG system to serve real people.
Over several months, I built Emerald City Beacon, a dual-platform AI solution supporting marginalized communities in King County. The architecture is split into two core apps:
- Django: Handles structured data ops—ingestion, validation, conflict resolution, and human override. It runs nightly pipelines across sources like WA 211, local food banks, and internal spreadsheets.
- FastAPI: A standalone RAG backend serving vector search queries via OpenAI + Pinecone. It’s wrapped in a minimal HTML frontend and tuned for accessibility and screen-reader use.
Together, they form a trusted, human-centered resource finder. Real-time answers, no hallucinations, no paywall. Everything is open format, auditable, and runs clean Python. This is what it looks like to build AI that actually works for the people who need it most.
