Hi👋 I'm Mohamed Sobhy!

I am an MPhil student in Computer Science at the Hong Kong University of Science and Technology (HKUST), advised by Prof. Qifeng Chen.

My research focuses on robot learning and embodied intelligence, particularly humanoid loco-manipulation, learning from human demonstrations, and building systems that work on real robots.

I currently work with the HKUST Robotics Institute on:

• Converting monocular human videos into whole-body humanoid control policies
• Teleoperation and imitation learning for multi-step humanoid manipulation
• Spatial memory and mapping for locomotion in unstructured environments
• Dexterous Manipulation

My work spans human-motion recovery, motion retargeting, robot data collection, generative policies, simulation, and deployment on platforms including the Unitree G1.

Before moving into robotics, I worked on controllable language-model inference at ETH Zürich, diffusion and vision-language-action models at Upsurge Labs, and machine-learning systems at HKUST and Ansel Intelligence. I also spent several years competing in algorithmic programming, reaching Codeforces Master and earning two ICPC regional silver medals.

You can connect with me on:

• X: @mohamedsobhi777
• Github: @mohamedsobhi777
• LinkedIn: /mohamed-sobhy-morsi
• Email mohamedsobhi777@gmail.com

Other Pages

Mohamed Sobhy - Experience
Mohamed Sobhy - Projects

I am interested in research collaborations and internships involving robot learning, humanoid robotics, embodied AI, and foundation models for physical systems.