A collection of Retrieval-Augmented Generation (RAG) tutorials and examples using Qdrant vector database and modern AI models.

This repository contains three (for now) comprehensive RAG implementations, from basic document Q&A to advanced agentic workflows. Each tutorial is designed for different skill levels and use cases.

Complete beginner's guide to Qdrant vector database

Comprehensive tutorial materials for learning Qdrant from scratch, including setup guides and hands-on examples.

What's Included:

• Step-by-step installation (Docker and local)
• Interactive Jupyter notebook with real examples
• Performance insights and best practices
• AINS 3.0 conference materials

Location: ./AINS_qdrant_talk/

Best for: Vector database beginners, Qdrant evaluation, educational workshops

Beginner-friendly implementation for quick starts

A streamlined RAG chatbot that answers questions about your documents using Qdrant, Gemini AI, and Docling for document processing.

Key Features:

• Easy setup with minimal configuration
• Interactive chat interface
• Smart document chunking with BGE embeddings
• Support for PDF documents
• In-memory or persistent storage options

Location: ./qdrant_simple_rag/

Best for: Personal document assistant, learning RAG basics, proof of concepts

Advanced implementation with intelligent document processing

An intelligent document analysis system that combines PDF processing, vector search, and agentic AI workflows using LangGraph.

Key Features:

• GPT-4o vision for accurate text extraction from complex PDFs
• LangGraph-powered agent that makes smart retrieval decisions
• Automatic document relevance evaluation
• Query rewriting for improved results
• Qdrant vector database with sub-24ms query latency

Location: ./Agentic-Qdrant-RAG/

Best for: Research paper analysis, legal document review, complex document workflows

cd qdrant_simple_rag
pip install -r requirements.txt

# Add your API key to .env file
echo "GEMINI_API_KEY=your_api_key_here" > .env

# Run with your document
python main.py --document data/your_document.pdf

cd Agentic-Qdrant-RAG
pip install pypdfium2 backoff langchain-community langchain langchain-openai langgraph qdrant-client

# Set environment variables
export OPENAI_API_KEY="your-openai-key"
export QDRANT_API_KEY="your-qdrant-key"  # Optional for local setup

cd AINS_qdrant_talk
jupyter notebook Qdrant_talk.ipynb

Or run directly in Google Colab

• Vector Database: Qdrant for fast similarity search
• Language Models: GPT-4o, Google Gemini, Groq
• Document Processing: Docling, pypdfium2 for PDF handling
• Embeddings: BGE, OpenAI embeddings, Sentence Transformers
• Frameworks: LangChain, LangGraph for RAG workflows
• Python Libraries: FastEmbed, Transformers

├── Agentic-Qdrant-RAG/          # Advanced agentic RAG system
├── AINS_qdrant_talk/            # Qdrant tutorial and setup guide
├── qdrant_simple_rag/           # Simple RAG chatbot implementation
├── README.md

Required environment variables:

# For Gemini models
GEMINI_API_KEY=your_gemini_key

# For GPT-4o and OpenAI embeddings  
OPENAI_API_KEY=your_openai_key

# For Groq models
GROQ_API_KEY=your_groq_key

# Optional - for remote Qdrant
QDRANT_URL=your_qdrant_url
QDRANT_API_KEY=your_qdrant_key

Enterprise Applications:

• Legal document review and analysis
• Technical documentation Q&A
• Research paper analysis
• Customer support knowledge bases

Educational & Research:

• Academic paper analysis
• Interactive learning materials
• Literature review automation
• Cross-document research

Personal Productivity:

• Document assistant and note-taking
• Reading companion for books/articles
• Meeting notes summarization

• Why Qdrant Will Be Your Favorite Vector Database
• How I Built an Agentic RAG System with Qdrant

Will soon expanding these tutorials to cover more advanced techniques and integrations, including:

• Advanced RAG strategies (e.g., re-ranking, query expansion)
• Integration with other vector databases
• Deployment strategies for RAG systems

Stay tuned for updates!

Contributions are welcome! Please see CONTRIBUTING.md for guidelines on how to contribute to this project.

For questions or support, connect with Mohamed Arbi Nsibi on LinkedIn.

Special thanks to AINS 3.0 organizers and the open source community for making these tutorials possible.