Jinxin Liu

Decoupling motion retargeting into a privileged generator and an anticipatory executor enables physical humanoids to mirror agile, expressive human motions in real time despite severe latency and embodiment mismatches.

Uncovering critical security and privacy vulnerabilities in foundation-model-powered robots could redefine how we approach their deployment in real-world applications.

GLM-5 doesn't just code; it engineers, showcasing unprecedented capability in tackling end-to-end software engineering challenges.