This project implements a scalable GenAI pipeline that integrates RAG, multi-vector retrieval, and a modular retriever-store architecture to enable contextual Q&A over multimodal structured PDFs. It optimizes retrieval relevance and input summarization using state-of-the-art LLMs and best MLOps practices, supported by Flask backend.