This repository accompanies a Chinese online course on RAG system development, implementing a complete end-to-end retrieval-augmented generation system using DeepSeek. The project is organized into 10 modules covering data loading, document chunking, text embedding with HuggingFace/BGE, vector storage with Milvus/Chroma, pre-retrieval and post-retrieval optimization, answer generation, and system evaluation using RAGAS/TruLens. Advanced topics include Graph RAG and multi-agent architectures.