RAG Q&A

Document Q&A Bot that implements Retrieval-Augmented Generation (RAG). Users can upload PDF or text documents and ask questions about their content.

Features

• Upload PDF and text documents (max 10MB)
• Ask natural language questions about uploaded content
• Real-time Q&A with source citations
• Vector-based document similarity search
• DeepSeek LLM integration for responses
• OpenAI embeddings for document processing

Project Structure

/
├── backend/          # Go API server
│   ├── cmd/
│   │   └── main.go   # Application entry point
│   ├── internal/
│   │   ├── config/   # Configuration handling
│   │   ├── handlers/ # HTTP handlers
│   │   └── services/ # Business logic (RAG pipeline, document processing)
│   ├── pkg/
│   │   ├── types/      # Data structures
│   │   └── utils/      # Utilities
│   │   └── similarity/ # Similarity search algorithm
│   ├── go.mod
│   ├── Makefile
│   ├── .env.example
│   └── .gitignore
└── frontend/         # Next.js React application
    ├── src/
    │   ├── app/      # Next.js pages
    │   └── types/
    ├── package.json
    ├── tsconfig.json
    ├── .env.example
    └── .gitignore

Quick Start

Backend Setup (Go API)

cd backend

# Install dependencies
go mod download

# Configure environment
cp .env.example .env
# Edit .env with your API keys:
# DEEPSEEK_API_KEY=your_deepseek_api_key_here
# OPENAI_API_KEY=your_openai_api_key_here
# PORT=3001

# Start development server
go run cmd/main.go

The backend will run on http://localhost:3001

Frontend Setup (Next.js App)

cd frontend

# Install dependencies
npm install

# Configure environment
cp .env.example .env.local
# Edit .env.local with:
# NEXT_PUBLIC_BACKEND_URL=http://localhost:3001

# Start development server
npm run dev

The frontend will run on http://localhost:3000

API Endpoints

• POST /api/upload - Upload and process documents
• POST /api/query - Ask questions about uploaded documents
• GET /health - Health check

Environment Variables

Backend (.env)

• DEEPSEEK_API_KEY - DeepSeek Chat API key for LLM responses
• OPENAI_API_KEY - OpenAI API key for document embeddings
• PORT - Server port (default: 3001)

Frontend (.env)

• NEXT_PUBLIC_BACKEND_URL - Backend API URL (default: http://localhost:3001)

Technology Stack

Backend

• Go 1.24.4 - Programming language
• Gin - Web framework
• DeepSeek API - Language model for responses
• OpenAI Embeddings - Document vectorization
• ledongthuc/pdf - PDF text extraction
• In-memory Vector Store - Document similarity search

Frontend

• Next.js - React framework
• React - UI library
• TypeScript - Type safety
• Tailwind CSS - Styling

Architecture

RAG Pipeline

1. Document Upload: Files are processed and chunked into 1000-character segments with 200-character overlap
2. Embedding: Text chunks are converted to vectors using OpenAI embeddings
3. Storage: Vectors stored in memory (ephemeral - resets on restart)
4. Query: User questions trigger similarity search to find relevant chunks
5. Generation: DeepSeek LLM generates responses based on retrieved context

Data Flow

Frontend → Go Backend /api/upload → Document Processing → Vector Storage
Frontend → Go Backend /api/query → Similarity Search → DeepSeek LLM → Response