AnkurJain305/MultiModal-Rag- ? reverse-engineered prompt

Build me a Python app that does multimodal RAG for documents with both text and images, especially PDFs, scans, and files with diagrams or floorplans.

I want to be able to upload a document, extract the text and any images, break the text into overlapping chunks, create embeddings for both the text and images, and store everything in a vector database so I can search it later. Then when I ask a question, it should find the most relevant text and image pieces and use them to generate a helpful answer with context.

Please make it work with common tools like LangChain, Pillow, and a vector store such as FAISS or Chroma, and keep it easy to run in a notebook or a simple Python script. Use OpenAI or Hugging Face style embeddings if needed, and include OCR support for scanned pages if that helps. If you need to look up current docs online, go ahead.