Max Robotics — Global Service Robot Directory + US Certification

About

Rae Jeong is a robotics and machine learning researcher whose work sits at the intersection of reinforcement learning, sim-to-real transfer, and data-driven robot control. Their research tackles some of the most pressing challenges in making robotic systems practical and scalable in real-world environments. Jeong is perhaps best known for pioneering a framework for data-driven robotics that combines large-scale recorded robot experience with learned reward functions — an approach that has garnered over 100 citations across multiple publications and demonstrated success on real robot manipulation tasks. This work introduced "reward sketching" as an accessible means of specifying tasks without dense human supervision. Complementing this, their research on self-supervised sim-to-real adaptation (accumulating over 65 citations) offers elegant solutions for bridging the gap between simulation and reality using unlabeled real-world visual data. Jeong has also made notable contributions to robust reinforcement learning, developing frameworks that account for model misspecification in continuous control — a critical concern for deploying RL in uncertain real environments. Their work on dexterous manipulation, complex object stacking, and learning from suboptimal experts further demonstrates a commitment to advancing general-purpose robotic intelligence. Collectively, their publications reflect a rigorous and impact-driven research agenda that continues to shape modern robot learning.