Design Multimodal RAG for Text and Images | Dataford Interview Questions

Scenario

You are building a question answering feature for an internal knowledge base that contains both documents and images, such as screenshots, diagrams, charts, and scanned pages. Users ask natural language questions that may require evidence from either modality, or both. The team wants answers that are grounded in the source material and can point to the exact text or image region used.

Question

How would you handle multi-modal RAG with both text and images?

Problem

Scenario

Question

How would you handle multi-modal RAG with both text and images?

Problem

Scenario

Question

How would you handle multi-modal RAG with both text and images?

Problem

Scenario

Question

How would you handle multi-modal RAG with both text and images?