Agentic RAG

A small-scale PDF-based Retrieval-Augmented Generation (RAG) system designed for correctness, clear separation of concerns, and practical efficiency.

Features

• Persistent Vector Storage: Uses Pinecone for scalable vector storage
• Incremental Indexing: Selective reindexing for changed PDFs only
• Grounded Answer Generation: Pluggable LLM layer with answer grounding
• Local Caching: SQLite-backed caching to avoid unnecessary LLM calls
• CLI Interface: Command-line tools for indexing and querying
• FastAPI API: HTTP endpoints for programmatic access
• Analytics: Query logging for internal analytics

Architecture

The system follows a clean pipeline: PDF -> chunk -> embed -> Pinecone -> retrieve -> LLM

Components

1. Ingestion: Loads PDFs from data/, cleans text, splits into chunks
2. Embedding: Uses sentence-transformers/all-MiniLM-L6-v2 for embeddings
3. Vector Storage: Pinecone for persistent vector storage
4. Retrieval: Semantic search over embedded chunks
5. Generation: LLM-powered answer generation with grounding

Installation

1. Clone the repository:

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

2. Create 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 (Pinecone, Google AI, etc.)

Usage

CLI

Index documents:

python -m app.main index

Ask questions:

python -m app.main ask "What is the main topic of the documents?"

API

Start the FastAPI server:

uvicorn app.api:app --reload

The API will be available at http://localhost:8000

Configuration

The system uses pydantic-settings for configuration. Key settings include:

• Pinecone API key and environment
• Google AI API key for LLM
• Embedding model configuration
• Chunk size and overlap settings

See app/config.py for all available options.

Testing

Run tests:

pytest

Project Structure

app/
├── core/          # Core RAG pipeline components
├── infra/         # Infrastructure (embedding, storage, etc.)
├── utils/         # Utilities (logging, caching, etc.)
├── api.py         # FastAPI application
├── config.py      # Configuration management
├── graph.py       # LangGraph orchestration
└── main.py        # CLI entrypoint

tests/             # Test suite
data/              # PDF documents directory

Contributing

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Add tests if applicable
5. Submit a pull request

License

MIT License - see LICENSE file for details