RAG-based QA Bot

Built a RAG (Retrieval-Augmented Generation) pipeline using LangChain for orchestration, integrating Pinecone vector database for semantic document storage and retrieval, achieving high-relevance retrieval accuracy by combining dense vector search with keyword-based retrieval and re-ranking models.

Implemented context-aware answer generation with the Cohere API, working around context-window limits through chunking and hybrid search so the retrieved passages are the ones the question actually needs.

Developed a multi-document knowledge base system that brings information to life by connecting retrieval with generation, enabling accurate question answering through the fusion of retrieval and generation, where context meets intelligence to provide meaningful responses from document collections.