Anupam Rawat

Closing the Perception Gap: Architecting Multi-Modal Agents with Open-Vocabulary Alignment

As autonomous agents transition from text-based assistants to operational systems that navigate physical and digital environments, bridging vision and language becomes critical. This session presents a technical deep dive into resilient multi-modal agents using open-vocabulary knowledge transfer, including methods for stronger regional visual reasoning in out-of-distribution scenarios.