Azure Foundry + Postgres RAG demo

This demo shows a minimal RAG pipeline using an ASP.NET Web API, PostgreSQL, pgvector extension, and Azure Foundry client. You can read more about this demo here: https://agentic-ai.info/implementing-rag-with-azure-foundry-net-backend-and-postgresql-vector-database/

Quick start:

1. Build and start the stack:

docker compose up --build

2. Create an embedding (or skip and go to Point 4. Embed a local file):

curl -X POST http://localhost:5001/embeddings \
  -H 'Content-Type: application/json' \
  -d '{"text":"Hello from demo","metadata":{"source":"sample"}}'

3. Query:

curl -X POST http://localhost:5001/query \
  -H 'Content-Type: application/json' \
  -d '{"question":"What is Hello from demo?"}'

4. Embed a local file (use the demo text provided in demo-data.txt):

python3 scripts/embed_file.py scripts/demo-data.txt --source demo-data

Optional arguments:

python3 scripts/embed_file.py scripts/demo-data.txt --chunk-size 1200 --overlap 150 --api-base-url http://localhost:5001 --source demo-data

Example queries (after embedding demo-data.txt):

curl -X POST http://localhost:5001/query \
  -H 'Content-Type: application/json' \
  -d '{"question":"Which city did the narrator visit after Lisbon?"}'

curl -X POST http://localhost:5001/query \
  -H 'Content-Type: application/json' \
  -d '{"question":"What are the best month for travel?"}'

Notes:

• Configure Azure credentials in .env (copy .env.template as a starting point). The API will fail to start if any of the three required variables (AZURE_FOUNDRY_API_KEY, AZURE_FOUNDRY_ENDPOINT, AZURE_FOUNDRY_EMBEDDINGS_ENDPOINT) are missing.
• Vector search uses the pgvector extension with a vector(1536) column, matching the text-embedding-ada-002 / text-embedding-3-small output dimension. If you use different embedding models the db init script needs to be adjusted.