Deepak Pathak

Deepak Pathak is the Raj Reddy Assistant Professor in Robotics at Carnegie Mellon University's School of Computer Science, where he leads the Pathak Research Group in the Robotics Institute and is affiliated with the Machine Learning Department. His research focuses on building agents with human-like generalization capabilities through self-supervised learning from raw sensory data at the intersection of computer vision, machine learning, and robotics.

• PhD from UC Berkeley (advised by Alyosha Efros & Trevor Darrell)
• Bachelors in Computer Science from IIT Kanpur
• Research supported by Facebook, Nvidia, and Snapchat fellowships

His work spans robot hands, legged robots, world modeling, 3D vision, and human-centered robotics, with key contributions in sim2real transfer, test-time adaptation, and structured world models. Recent articles focus on force-attending policies, zero-shot manipulation, and generative classifiers.

Scientific accolades include:

• Sloan Research Fellowship 2025
• MIT TR 35 under 35 Innovator 2024
• Raj Reddy Chair at CMU 2024
• Okawa Research Award 2022
• Young Alumnus Award from IIT Kanpur 2022

He has advised 10+ PhD students and 15+ Masters students, with notable alumni now at top PhD programs (MIT, Stanford, Princeton) and industry (Tesla Optimus, Skild AI). His lab develops cost-effective hardware like LEAP Hand (2000 USD) and DASH (7000 USD) for dexterous manipulation research.