Jay Vakil

Papers

Total Citations

114

H-Index

About

Jay Vakil is an emerging researcher at the intersection of robotics, machine learning, and embodied artificial intelligence. His work focuses on developing scalable, generalizable frameworks that enable robots to operate effectively across diverse real-world environments — a challenge that sits at the heart of modern robotics research. Vakil's most impactful contribution, RoboAgent (2024, 70 citations), addresses the critical bottleneck of limited robotics training data by leveraging semantic augmentations and action chunking to build manipulation agents capable of generalizing across arbitrary objects and settings. Complementing this, his work on OK-Robot (2024, 41 citations) demonstrates how open-knowledge vision and language models can be practically integrated with navigation and grasping systems to create capable, deployable robotic assistants. Together, these papers reflect his commitment to bridging cutting-edge AI with real-world robotic utility. His contribution to RoboHive (2023), a unified software ecosystem for robot learning research, further illustrates his investment in building infrastructure that benefits the broader research community. Spanning dexterous manipulation to whole-arm control, RoboHive lowers barriers for researchers entering the field. With over 110 cumulative citations, Vakil is establishing himself as a thoughtful and productive voice in the rapidly advancing field of robot learning.