Personal RAG Application (In-development)

A FastAPI-based Retrieval-Augmented Generation (RAG) application that integrates multiple AI services and vector embeddings for document processing and AI interactions.

Technology Stack

Core Frameworks & Libraries

• FastAPI: Main web framework for building the API
• Pydantic: Data validation using Python type annotations
• Uvicorn: ASGI server implementation
• httpx: Async HTTP client for making API calls
• Rich: Advanced terminal output formatting and logging

AI & ML Integration

• HuggingFace: Integration for transformer models and embeddings
  • Support for both API and local model implementations
  • Sentence transformers for embeddings
• OpenRouter: Multi-model AI platform integration
  • Access to various LLMs (Claude-3, GPT-4, Gemini Pro, Mistral, Qwen)
  • Embeddings generation capabilities

Database & Authentication

• Supabase:
  • Authentication and user management
  • Vector storage and similarity search
  • Token management

Document Processing

• MarkItDown: PDF parsing and text extraction
• Support for both PDF and text file processing

Payment Integration

• LemonSqueezy: Payment processing and subscription management
  • Token-based usage tracking
  • Webhook handling for subscription events

Key Features

• User authentication and authorization
• File upload and processing (PDF and text)
• Vector embeddings generation and storage
• Token-based access control
• AI model interaction with multiple providers
• Subscription management
• Rich logging and error handling

Authentication & Security

• JWT-based authentication
• Token balance verification
• Webhook signature verification
• Rate limiting and usage tracking

Project Structure

.
├── api/
│   ├── core/
│   │   ├── logger.py
│   │   └── parser.py
│   ├── services/
│   │   ├── huggingface/
│   │   ├── openrouter/
│   │   ├── supabase/
│   │   └── lemonsqueezy/
│   ├── middlewares/
│   └── routers/