A minimal Retrieval-Augmented Generation (RAG) pipeline using OpenAI and ChromaDB to answer questions from your own PDF documents.

• Initialize OpenAI embeddings and chat models
• Ingest PDF documents and extract text
• Split text into overlapping chunks
• Store embeddings in a persistent ChromaDB collection
• Load ChromaDB and retrieve relevant chunks
• Generate responses to user questions using OpenAI Chat API
• .env support with .gitignore to keep secrets out of version control

1. Create and activate a virtual environment (optional but recommended).
2. Install dependencies: pip install -r requirements.txt
3. Create a .env file in the project root with your OpenAI API key: OPENAI_API_KEY=sk-...

   Backward compatibility if you already have this:

   OPENAI_KEY=sk-...

Optional environment variables:

• OPENAI_EMBEDDING_MODEL (default: text-embedding-3-small)
• OPENAI_CHAT_MODEL (default: gpt-4o-mini)
• CHROMA_DIR (default: ./chroma)
• CHROMA_COLLECTION (default: rag_docs)
• CHUNK_SIZE (default: 1000)
• CHUNK_OVERLAP (default: 200)
• TOP_K (default: 3)

Ingest PDFs from a folder: python -m src.main ingest path\to\pdf_folder

Ask a question: python -m src.main ask "What are the key points?"

• The Chroma vector store is persisted under CHROMA_DIR, which you may want to add to .gitignore if committing publicly.
• Ensure your .env is listed in .gitignore (it is included by default in this repo).