AI Document Analysis Agent

About the Project

Architected an autonomous, agentic RAG (Retrieval-Augmented Generation) workflow implementing complex query decomposition and multi-modal knowledge retrieval from unstructured PDF documents. Engineered a sophisticated pipeline integrating OpenCV for computer vision-based document parsing, LangChain for orchestration, and local LLM inference via Ollama. Implemented vector similarity search using ChromaDB/Qdrant for semantic indexing, with Docker-containerized FastAPI endpoints providing RESTful API access. The system demonstrates advanced NLP techniques including document chunking, embedding generation, and context-aware retrieval for enterprise-grade document intelligence applications.