My Mission
My mission is to enable safe and reliable humanoid autonomy in human environments for the benefit of humanity, and to leave the world better than I found it. Recent progress with foundation models suggests that we’re already close to achieving general humanoid autonomy. However, current real-world deployment will rely on human supervision, similar to safety teleoperation in autonomous driving. Despite massive long-term investments (Google/Waymo invested ~$100 billion in the last 15 years), fully unsupervised autonomy has not been achieved.
As Ilya Sutskever remarked, we’re shifting from the “age of scaling” back into an “age of research”. Simply investing more compute or capital is not enough. Instead, we need research on fundamental algorithmic breakthroughs.
To achieve my dream, I’m tackling the following subproblems:
- Safety and reliability guarantees with safe control methods
- Unify manipulation and locomotion in one safe loco-manipulation framework
- Semantic reasoning with foundation models
- Testing safe humanoids in real-world industrial and domestic use cases.
In parallel, I’m interested in deeper theoretical questions that may inform long-term humanoid intelligence:
- Algorithmic foundations of AI agents with neuroscience insights, e.g., continual learning with scalable brain-inspired architectures
- Mathematical theory of AI
Overall, my work spans the full technology pipeline of intelligent humanoid systems:
If you are interested in collaborating on any of these topics, please don’t hesitate to contact me via email or LinkedIn!