All projects

    P-014 · Case study

    RESEARCH

    AI Document Analysis Agent

    An agentic RAG workflow that breaks complex questions into sub-queries and pulls multi-modal answers out of unstructured PDFs. Inference runs locally via Ollama.

    PythonLangChainOllamaChromaDBQdrantFastAPIDockerOpenCVRAG
    P-014 / Register plateRESEARCH
    No plate photograph on file · schematic record only
    PythonLangChainOllama

    More detail

    An agentic RAG workflow for complex query decomposition and multi-modal knowledge retrieval from unstructured PDFs. OpenCV handles computer-vision document parsing, LangChain handles orchestration, and LLM inference runs locally via Ollama. ChromaDB and Qdrant back the vector similarity search; Docker-containerized FastAPI endpoints expose it all over REST.