Design a multimodal RAG assistant

Design a retrieval-augmented generation (RAG) service that answers natural-language questions over an internal multimodal knowledge base. The knowledge base contains at least text and images, and may also include PDFs or tabular data. Your system must retrieve relevant evidence from multiple modalities and use a language model to generate an answer that is grounded in that evidence and includes citations. The available content may include: text documents such as wiki pages…