🤖 Chat with PDF locally using Ollama + LangChain

A powerful local RAG (Retrieval Augmented Generation) application that lets you chat with your PDF documents using Ollama and LangChain. This project includes multiple interfaces: a modern Next.js web app, a Streamlit interface, and Jupyter notebooks for experimentation.

✨ Features

• 🔒 100% Local - All processing happens on your machine, no data leaves
• 📄 Multi-PDF Support - Upload and query across multiple documents
• 🧠 Multi-Query RAG - Intelligent retrieval with source citations
• 🎯 Advanced RAG - LangChain-powered pipeline with ChromaDB
• 🖥️ Two Modern UIs - Next.js (primary) and Streamlit interfaces
• 🔌 REST API - FastAPI backend for programmatic access
• 📓 Jupyter Notebooks - For experimentation and learning

🖼️ Screenshots

Next.js Interface (Recommended)

Modern chat interface with PDF management, source citations, and reasoning steps

Streamlit Interface

Classic Streamlit interface with PDF viewer and chat functionality

📺 Video Tutorial

🏗️ Project Structure

ollama_pdf_rag/
├── src/
│   ├── api/                  # FastAPI REST API
│   │   ├── routers/          # API endpoints
│   │   ├── services/         # Business logic
│   │   └── main.py           # API entry point
│   ├── app/                  # Streamlit application
│   │   ├── components/       # UI components
│   │   └── main.py           # Streamlit entry point
│   └── core/                 # Core RAG functionality
│       ├── document.py       # PDF processing
│       ├── embeddings.py     # Vector embeddings
│       ├── llm.py            # LLM configuration
│       └── rag.py            # RAG pipeline
├── web-ui/                   # Next.js frontend
│   ├── app/                  # Next.js app router
│   ├── components/           # React components
│   └── lib/                  # Utilities & AI integration
├── data/
│   ├── pdfs/                 # PDF storage
│   └── vectors/              # ChromaDB storage
├── notebooks/                # Jupyter notebooks
├── tests/                    # Unit tests
├── docs/                     # Documentation
├── run.py                    # Streamlit runner
├── run_api.py                # FastAPI runner
└── start_all.sh              # Start all services

🚀 Getting Started

Prerequisites

1. Install Ollama

   • Visit Ollama's website to download and install
   • Pull required models:

     ollama pull llama3.2  # or your preferred chat model
     ollama pull nomic-embed-text  # for embeddings
     

2. Clone Repository

   git clone https://github.com/tonykipkemboi/ollama_pdf_rag.git
   cd ollama_pdf_rag
   

3. Set Up Python Environment

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

4. Set Up Next.js Frontend (for the modern UI)

   cd web-ui
   pnpm install
   pnpm db:migrate
   cd ..
   

🎮 Running the Application

Option 1: Next.js + FastAPI (Recommended)

Start both services:

# Terminal 1: Start the FastAPI backend
python run_api.py
# Runs on http://localhost:8001

# Terminal 2: Start the Next.js frontend
cd web-ui && pnpm dev
# Runs on http://localhost:3000

Or use the convenience script:

./start_all.sh

Service URLs:

Option 2: Streamlit Interface

python run.py
# Runs on http://localhost:8501

Option 3: Jupyter Notebook

jupyter notebook

Open notebooks/experiments/updated_rag_notebook.ipynb to experiment with the code.

💡 Usage

Next.js Interface

1. Upload PDFs - Click the 📎 button or drag & drop files
2. View PDFs - Uploaded PDFs appear in the sidebar with chunk counts
3. Select Model - Choose from your locally available Ollama models
4. Ask Questions - Type your question and get answers with source citations
5. View Reasoning - See the AI's thinking process and retrieved chunks

Streamlit Interface

1. Upload PDF - Use the file uploader or toggle "Use sample PDF"
2. Select Model - Choose from available Ollama models
3. Ask Questions - Chat with your PDF through the interface
4. Adjust Display - Use the zoom slider for PDF visibility
5. Clean Up - Delete collections when switching documents

🔌 API Reference

The FastAPI backend provides these endpoints:

See full documentation at http://localhost:8001/docs when running.

🧪 Testing

# Run all tests
python -m pytest tests/ -v

# Run with coverage
python -m pytest tests/ --cov=src

Pre-commit Hooks

pip install pre-commit
pre-commit install

⚠️ Troubleshooting

• Ollama not responding: Ensure Ollama is running (ollama serve)
• Model not found: Pull models with ollama pull <model-name>
• No chunks retrieved: Re-upload PDFs to rebuild the vector database
• Port conflicts: Check if ports 3000, 8001, or 8501 are in use

Common Errors

ONNX DLL Error (Windows)

DLL load failed while importing onnx_copy2py_export

Install Microsoft Visual C++ Redistributable and restart.

CPU-Only Systems

Reduce chunk size if experiencing memory issues:

• Modify chunk_size to 500-1000 in src/core/document.py

🤝 Contributing

• Open issues for bugs or suggestions
• Submit pull requests
• Comment on the YouTube video for questions
• ⭐ Star the repository if you find it useful!

📝 License

This project is open source and available under the MIT License.

⭐️ Star History

Built with ❤️ by Tony Kipkemboi

Follow me on X | LinkedIn | YouTube | GitHub