I build high-performance, scalable systems that connect generative worlds, hybrid simulation, and Sim2Real learning, enabling embodied agents to continually learn from their own rollouts in both simulated and real-world environments.
- Systems: high-performance, heterogeneous, GPU-accelerated simulation and scalable data-generation and training infrastructure
- Simulation: generative, neural, and differentiable simulation for robot learning
- Embodied intelligence: physics-structured models, Sim2Real transfer, and online and continual learning
EmbodiChain is an open-source, end-to-end, GPU-accelerated framework that turns this research agenda—scalable systems, generative and differentiable simulation, and continual Sim2Real learning—into practical infrastructure for simulation, automated data generation, robot learning, and real-world deployment. The project is under active development, and I welcome researchers, engineers, and open-source contributors interested in new robots, tasks, simulation capabilities, learning methods, or real-world applications to collaborate and build it together.
For research collaborations, student internships, open-source projects, or industry partnerships beyond EmbodiChain, please feel free to get in touch.