Agentic RAG System with Multi-Agent Orchestration

An end-to-end Retrieval-Augmented Generation (RAG) system that leverages multiple intelligent agents for enhanced document understanding, retrieval, and response generation with comprehensive source attribution.

🌟 Features

• Multi-Format Document Processing

  • PDF, DOCX, HTML support
  • Hierarchical document parsing
  • Metadata extraction
  • Page-level information preservation

• Intelligent Agent System

  • Retriever Agent: Smart document chunk retrieval
  • Metadata Agent: Context-aware filtering
  • Reflection Agent: Response quality assurance
  • Web Search Agent: External knowledge integration

• Advanced RAG Capabilities

  • Hybrid search (semantic + keyword + metadata)
  • Composable graph indices
  • Source attribution tracking
  • Confidence scoring

🏗️ Project Structure

agentic_rag/
├── app/                    # Application code
│   ├── api/               # FastAPI endpoints
│   ├── core/              # Core business logic
│   │   ├── agents/        # Agent implementations
│   │   ├── parsers/       # Document parsers
│   │   └── retrievers/    # Retrieval logic
│   ├── models/            # Data models
│   └── utils/             # Utility functions
├── config/                # Configuration files
├── data/                  # Data storage
│   ├── documents/         # Uploaded documents
│   └── vector_store/      # Vector database
├── tests/                 # Test suite
└── scripts/               # Utility scripts

🚀 Getting Started

Prerequisites

• Python 3.9+
• Docker (optional, for containerized deployment)
• LLAMA API key (for LLAMAParse)

Installation

1. Clone the repository:

   git clone https://github.com/yourusername/agentic-rag.git
   cd agentic-rag

2. Create and activate a virtual environment:

   python -m venv venv
   source venv/bin/activate  # On Windows: .\venv\Scripts\activate

3. Install dependencies:

   pip install -r requirements.txt

4. Set up environment variables:

   cp .env.example .env
   # Edit .env with your API keys and configuration

Running the System

1. Start the vector database:

   docker-compose up -d qdrant

2. Launch the API server:

   uvicorn app.api.main:app --reload

3. Access the API documentation at http://localhost:8000/docs

🛠️ Usage

Document Ingestion

from app.core.parsers import DocumentParser

parser = DocumentParser()
document = parser.parse("path/to/document.pdf")

Query Processing

from app.core.agents import QueryProcessor

processor = QueryProcessor()
response = processor.process_query(
    "What was the product launch date?",
    metadata_filters={"date_range": "2023"}
)

Example Response

{
    "answer": "The product was launched in March 2023.",
    "sources": [
        {
            "type": "document",
            "document": "Product_Overview.pdf",
            "page": 14,
            "chunk_id": "c_004",
            "heading": "Product Launch Details",
            "confidence": 0.95
        }
    ]
}

📊 Evaluation

The system is evaluated on multiple metrics:

• Retrieval accuracy (Precision@K)
• Answer quality (BLEU/ROUGE + human evaluation)
• Reflection agent correction rate
• Response latency
• Source attribution correctness

🤝 Contributing

1. Fork the repository
2. Create a feature branch
3. Commit your changes
4. Push to the branch
5. Create a Pull Request

📝 License

This project is licensed under the MIT License - see the LICENSE file for details.

🙏 Acknowledgments

• LLAMAIndex team for the excellent RAG framework
• Qdrant team for the vector database
• All contributors and users of this project