datadays

Dockerized medical RAG app with:

• FastAPI backend for retrieval, reranking, and answer generation
• Next.js frontend with a polished query workspace
• one Docker service that runs both app processes in the same container

App layout

Backend

The backend wraps the existing prototype behavior behind API endpoints:

• GET /api/health
• GET /api/examples
• POST /api/query

It uses the shared RAG implementation in rag_query.py, so the CLI and API use the same retrieval and answer generation path.

Local Python dependency management is handled with uv.

Frontend

The frontend is a Next.js app-router interface built for:

• submitting natural-language medical queries
• showing health and collection state
• rendering structured answers
• exposing source document snippets

It proxies browser requests to the internal FastAPI process, so the browser only talks to one app origin.

Run with Docker

1. Copy .env.example to .env and fill in your real credentials.
2. Start the stack:

docker compose up --build

3. Open:

• app: http://localhost:3000
• backend docs through the container network: http://localhost:3000 for the UI, internal FastAPI stays on port 8000

Backend with uv

From the backend folder:

uv sync --dev
uv run pytest
uv run uvicorn app.main:app --reload

Repo layout

backend/        FastAPI API layer
frontend/       Next.js UI and API proxy routes
memory_bank/    Project context and working notes
rag_query.py    Shared RAG service used by CLI and backend
upload_data.py  Original upload script

Notes

• docker compose up --build now starts a single containerized app service.
• The backend defaults to drug_docs_v1, but you can override that with QDRANT_COLLECTION.
• If your Qdrant collection uses a different name, set it in .env before starting Docker.