An Intelligent PDF Analysis and Chat System powered by LLM + RAG

A production-ready Retrieval-Augmented Generation (RAG) system with a Flask web interface.

✅ Web Interface - Interactive Flask-based chat UI
✅ Document Upload - Easy PDF upload and processing
✅ Real-time Chat - Ask questions and get AI responses
✅ Semantic Search - Vector similarity matching
✅ Source Tracking - Shows document sources
✅ Database Management - Clear and manage documents
✅ Docker Ready - Production deployment
✅ Modular Architecture - Clean, extensible codebase

git clone https://github.com/askchandan/Project-Lexora.git
cd Project-Lexora

python -m venv venv
# Windows: venv\Scripts\activate
# macOS/Linux: source venv/bin/activate
pip install -r requirements.txt

OPENAI_API_KEY=your_api_key
OPENAI_API_BASE=https://openrouter.ai/api/v1
DATA_PATH=data
CHROMA_PATH=chroma_db
MODEL_NAME=mistralai/mistral-7b-instruct

python app.py

Visit http://localhost:5000

docker build -t project-lexora .
docker run -p 5000:5000 \
  -e OPENAI_API_KEY=your_key \
  -e OPENAI_API_BASE=https://openrouter.ai/api/v1 \
  -v lexora_data:/app/chroma_db \
  project-lexora

# Update .env with your API key
docker-compose up -d
docker-compose logs -f
docker-compose down

Visit http://localhost:5000

1. Click "Upload PDF" button
2. Select PDF file
3. Click "Upload"
4. Wait for success message

1. Type question in input box
2. Click "Send" or press Enter
3. View AI response with sources

1. Click "Clear Database"
2. Confirm action
3. Database resets

project-lexora/
├── app.py                  # Flask application
├── requirements.txt        # Dependencies
├── Dockerfile             # Docker image
├── docker-compose.yml     # Compose config
├── .env.example          # Env template
├── README.md             # This file
├── LICENSE               # MIT License
│
├── src/
│   ├── core/             # RAG pipeline
│   ├── models/           # LLM & Embeddings
│   ├── database/         # Vector store
│   └── utils/            # Config & logging
│
├── templates/
│   └── index.html        # Web UI
│
├── static/
│   ├── script.js         # Frontend logic
│   └── style.css         # Styling
│
├── data/                 # PDF storage
└── chroma_db/           # Vector database

# Required
OPENAI_API_KEY=sk_...
OPENAI_API_BASE=https://openrouter.ai/api/v1

# Optional
DATA_PATH=data
CHROMA_PATH=chroma_db
MODEL_NAME=mistralai/mistral-7b-instruct
TEMPERATURE=0.7
MAX_TOKENS=500

• mistralai/mistral-7b-instruct (default)
• gpt-3.5-turbo
• gpt-4
• Other OpenRouter models

Local

• Python 3.8+
• 2GB RAM
• 500MB disk

Docker

• Docker 20.10+
• Docker Compose 1.29+
• 2GB RAM
• 500MB disk

• PDF Processing: ~100ms per document
• Query Response: 2-5 seconds
• Search Speed: <50ms

✅ API keys in .env (gitignored)
✅ Input validation
✅ Secure storage
✅ CSRF protection ready

• Backend: Flask, LangChain, Chroma
• Frontend: HTML5, CSS3, JavaScript
• DevOps: Docker, Docker Compose
• AI: Mistral-7B via OpenRouter

1. Fork repository
2. Create feature branch
3. Make changes
4. Submit pull request

MIT License - see LICENSE file

Chandan Malakar
GitHub: @askchandan

Project Lexora - Intelligent RAG Chatbot
Powered by LangChain, Flask, and Chroma