rag-study-buddy

RAG Study Buddy is a local Retrieval-Augmented Generation (RAG) application for asking questions over PDF study notes. It uses Sentence Transformers, ChromaDB, and Ollama. The project was built as a hands-on learning project to understand the individual components of a RAG pipeline, including document processing, chunking, embeddings, vector retrieval, and grounded LLM generation. It allows a user to add PDF study material, extract and chunk document text, generate semantic embeddings, store embeddings in a vector database, retrieve relevant chunks for a question, generate answers using a locally hosted LLM, and view the document chunks used during retrieval. The pipeline loads a PDF with PyPDF, divides extracted text into overlapping chunks, converts each chunk into a dense vector using all-MiniLM-L6-v2 from Sentence Transformers, and stores embeddings, original text, and source metadata in ChromaDB. For a question, the system embeds the question with the same model, performs vector similarity search, adds retrieved chunks to the prompt, and sends it to a locally running Llama 3.2 model through Ollama. The model is instructed to answer only using the supplied context. Both the vector database and the LLM run locally, and the project does not require a hosted LLM API for question answering.