The student is expected to work on novel methods for reasoning in large multimodal models, with a focus on enabling multimodal systems to understand, interpret, and reason over visual and textual information jointly. Research topics may include compositional reasoning, grounded inference, multimodal chain-of-thought methods, evaluation benchmarks, and trustworthy AI. The project aims to advance the capabilities of next-generation AI and robotics systems for complex real-world tasks requiring robust multimodal understanding and reasoning.