This project is a demo of a simple Retrieval-Augmented Generation (RAG) system, featuring a FastAPI backend with document ingestion and querying, and a React + Vite frontend for user interaction.

• Backend (FastAPI, ChromaDB, Transformers):

  • Ingest plain text documents into a vector database (ChromaDB)
  • Query the database for relevant documents using semantic search
  • Generate answers to user questions using a GPT-2 model, augmented with retrieved context

• Frontend (React, Vite):

  • Upload and ingest text files
  • Ask questions and receive answers with supporting context
  • Simple, modern UI for demo purposes

backend/ main.py # FastAPI app, ChromaDB, GPT-2 integration send_txt_to_api.py # CLI utility to ingest .txt files via API requirements.txt # Python dependencies frontend/ src/App.tsx # Main React app (UI for ingest/query) index.html # HTML entry point package.json # Frontend dependencies vite.config.mts # Vite config 

1. Install Python dependencies:

   cd backend pip install -r requirements.txt

2. Run the FastAPI server:

   uvicorn main:app --reload

   The API will be available at http://localhost:8000.

3. (Optional) Ingest .txt files via CLI:

   python send_txt_to_api.py <directory_with_txt_files> # Or specify a custom API URL: python send_txt_to_api.py <dir> --api-url http://localhost:8000/ingest

1. Install Node dependencies:

   cd frontend npm install

2. Start the development server:

   npm run dev

   The app will be available at http://localhost:3000.

1. Ingest Documents:

   • Use the web UI to upload .txt files, then click "Ingest".
   • Or use the CLI utility to batch-ingest files.

2. Ask Questions:

   • Enter a question in the web UI and click "Query".
   • The backend retrieves relevant documents and generates an answer using GPT-2.

• Backend: FastAPI, ChromaDB, Transformers (GPT-2), Torch
• Frontend: React, Vite, TypeScript

• This demo uses GPT-2 for simplicity; for production, use a more capable model.
• ChromaDB is used for vector search; you can swap for other vector DBs as needed.