DeadRAG 🧠💬

Note: Project is neither active in development at the moment nor finished

An intelligent RAG (Retrieval-Augmented Generation) chat application. Built with a modular architecture, DeadRAG processes PDF documents with advanced image extraction, vectorization, and AI-powered conversations about both text and visual content.

✨ Features

📄 Advanced PDF Processing

• Upload and intelligently process PDF documents
• Image Extraction & Analysis - Automatically extracts images from PDFs and generates AI descriptions
• Multi-Modal Vectorization - Both text content and image descriptions are vectorized for comprehensive search
• Visual Question Answering - Ask questions about charts, diagrams, or any images within your PDFs

🧩 Modular & Extensible Architecture

• Pluggable LLM Providers - Currently uses Anthropic Claude, easily extendable to other providers
• Flexible Ingestion Pipeline - Modular design allows for different document processing methods
• Provider-Agnostic Design - Swap components without major refactoring

💬 Intelligent Chat Experience

• Real-time Communication - WebSocket-powered chat for instant responses
• Context-Aware Responses - AI understands both textual and visual content from your documents
• Multi-Modal Queries - Ask about text, images, or combinations of both

🔧 Developer Experience

• 📊 LLM Observability - Integrated Langfuse for monitoring AI interactions and performance
• 🎨 Modern UI/UX - Beautiful, responsive interface built with Nuxt 3 and TailwindCSS
• 🐳 Container-Ready - Fully dockerized external services for seamless deployment
• 🔒 Type-Safe - Complete TypeScript implementation across the entire stack

🏗️ Tech Stack

Frontend

• Framework: Nuxt 3 with Vue 3
• Styling: TailwindCSS + Nuxt UI
• State Management: Pinia
• Real-time: Socket.io Client

Backend

• Framework: NestJS
• Database: PostgreSQL with Drizzle ORM
• AI/LLM: Anthropic Claude
• WebSockets: Socket.io

External Services

• Document Processing: Unstructured API
• LLM Monitoring: Langfuse

🚀 Quick Start

Prerequisites

• Node.js (v18 or higher)
• Docker & Docker Compose
• PostgreSQL database

Installation

1. Start external services

   docker-compose up -d

2. Install dependencies

   # Install frontend dependencies
   cd frontend
   npm install
   
   # Install backend dependencies
   cd ../backend
   npm install

3. Environment Setup

   Create environment files for both frontend and backend:

   Examples are included in the frontend/backend folder

4. Start the development servers

   In separate terminals:

   # Start docker containers (runs unstructured and langfuse)
   docker compose up
   
   # Start backend (runs on http://localhost:3010)
   cd backend
   npm run start:dev
   
   # Start frontend (runs on http://localhost:3000)
   cd frontend
   npm run dev

🛠️ Development

Project Structure

dead-rag/
├── frontend/              # Nuxt 3 frontend application
│   ├── components/        # Vue components
│   ├── pages/            # Application pages
│   ├── stores/           # Pinia stores
│   └── types/            # TypeScript type definitions
├── backend/              # NestJS backend application
│   ├── src/
│   │   ├── chat/         # Chat functionality
│   │   ├── documents/    # Document management
│   │   ├── ingestion/    # Document processing
│   │   ├── llm/          # LLM integration
│   │   └── database/     # Database configuration
├── docs/                 # Documentation
└── docker-compose.yml    # Docker services

External Services

1. Unstructured API: Runs on port 5001
2. Langfuse: Runs on port 5002
3. PostgreSQL: Configure connection in environment variables