This repository provides hands-on implementations of multiple RAG techniques including semantic chunking, context enrichment, document augmentation, query transformation, and reranking. It avoids heavy frameworks like LangChain or FAISS, instead using direct Python libraries (openai, numpy, fitz/pymupdf) for educational clarity. Each notebook is designed to be readable, modifiable, and runnable, serving as a practical learning resource for understanding how RAG systems work.