Modular RAG
Exposes a pluggable, observable, and modular Retrieval Augmented Generation (RAG) service framework through the Model Context Protocol for direct invocation by AI assistants.
Exposes a pluggable, observable, and modular Retrieval Augmented Generation (RAG) service framework through the Model Context Protocol for direct invocation by AI assistants.
The Modular RAG system provides a comprehensive, runnable engineering project that integrates key RAG components like hybrid search (dense + sparse with reranking), multi-modal visual processing (image captioning), and RAG evaluation, all exposed through the Model Context Protocol (MCP) for seamless integration with AI assistants like Copilot or Claude. Designed with a full-chain pluggable architecture, it allows for easy adaptation to diverse business needs by enabling 'Lego-block' style replacement of core modules, making it a highly flexible foundation for both learning RAG principles and building custom AI-driven applications. It also serves as a practical project and learning resource for those in large model-related roles.