mustapha muhsin

RAG Chatbot API

public code/independent project // backend & AI
PythonFastAPIPostgreSQLCeleryRedisPineconeAWS S3Docker

the idea

I built an API that accepts documents and answers questions using retrieved passages from those documents. The responses include source references so a user can trace an answer back to its material.

what i built

The FastAPI service handles uploads, document listing, chat, and conversation history. Uploaded files are stored in S3; PostgreSQL holds document and conversation data. A Celery worker processes ingestion in the background, with Redis as the task broker.

how it works

Documents are split into chunks, embedded, and indexed in Pinecone. At question time, the API retrieves relevant chunks and uses them to generate a grounded answer. The repository includes Docker setup, database migrations, and automated checks.