Multi-Modal Research Assistant

Design a research assistant that can process and retrieve information from both text and images in response to user queries. The assistant should allow users to upload images alongside text prompts and return relevant research articles, images, and summaries. Describe how you would handle data ingestion, embedding generation for both modalities, interaction with an LLM, and ensure accuracy and contextual relevance in responses.