Zhongang Cai
Staff Algorithm Scientist at SenseTime 商汤科技
Singapore
Summary
🤩
Rockstar🎓
Top SchoolZhongang Cai is a Staff Algorithm Scientist at SenseTime Research in Singapore with six years of experience bridging multimodal perception, 3D human reconstruction, and embodied AI. He leads teams building LLM-powered 3D autonomous characters and social agents grounded in research-grade human motion capture and generation. His publications include BiPointNet (ICLR'21), MLC-Net (ICCV'21) and MessyTable (ECCV'20), reflecting deep expertise across point clouds, multi-camera association, and model quantization. An active contributor to the OpenMMLab mmhuman3d toolbox, he implemented and refactored SMPLify components including camera modeling and 2D keypoint loss support, showing strong practical command of parametric human models. Zhongang holds a PhD from NTU and blends academic rigor with hands-on engineering—an example being a two-stage cascaded network enabling 2D CNNs for 3D tasks during his earlier NTU research. He’s notable for pairing foundational robotics/aerospace training with cutting-edge multimodal AI to push 3D social intelligence toward deployable agents.
6 years of coding experience
4 years of employment as a software developer
Singapore Universities Student Exchange Programme, Singapore Universities Student Exchange Programme at National University of SingaporeHigh School Science Stream, High School Science Stream at Anglo-Chinese Junior CollegeDoctor of Philosophy - PhD Computer Science, Doctor of Philosophy - PhD Computer Science at Nanyang Technological University Singapore
English, Chinese
Github Skills (4)
machine-learning10
computer-vision10
pytorch10
python9
Programming languages (2)
Github contributions (5)
open-mmlab/mmhuman3d
Jul 2021 - Dec 2022
OpenMMLab 3D Human Parametric Model Toolbox and Benchmark
Role in this project:
ML Engineer
ML EngineerContributions:1 release, 223 reviews, 94 commits in 1 year 5 months
Contributions summary:Zhongang contributed significantly to the `mmhuman3d` repository, which focuses on 3D human parametric models and related tools. Their commits primarily involved the development of `SMPLify`, including refactoring and adding features. These included the implementation of the camera module, and enhancements such as the support for 2D keypoint loss, and the refinement of the core functionality. The contributions demonstrated a strong understanding of the SMPL model and its integration.
3dopenmmlab
MotrixLab/SMPLer-X
Jun 2023 - Jan 2026
[NeurIPS 2023] Official Code for "SMPLer-X: Scaling Up Expressive Human Pose and Shape Estimation"
Contributions:3 reviews, 2 PRs, 20 pushes in 2 years 7 months
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.