Yu-te Cheng
Staff Engineer at XPENG
California, United States
Summary
🤩
Rockstar🎓
Top SchoolYu-te Cheng is a Staff Engineer based in California with 11 years of experience building production-grade deep learning systems for autonomous driving. Currently at XPENG, he leads end-to-end AI model development across 3D/2D BEV, vision-language models, and planning stacks, bridging perception and control. Previously at NVIDIA he shipped deep learning software for autonomous driving and contributed to the pytorch/tensorrt project by adding key PyTorch ops (e.g., aten::transpose, aten::layer_norm) to accelerate models on NVIDIA GPUs. Trained at Carnegie Mellon in robotics and computer vision, he combines research rigor with production engineering, often focusing on model-compiler and deployment robustness that quietly reduce failure modes in real-world driving stacks.
11 years of coding experience
5 years of employment as a software developer
Master's degree Robotics & Computer Vision, Master's degree Robotics & Computer Vision at Carnegie Mellon University
English, Japanese, Chinese
Github Skills (10)
tensorrt10
cuda10
pytorch10
machine-learning10
deeplearning-ai10
c-language10
deep-learning10
cprogramming-language10
nvidia9
libtorch9
Programming languages (1)
Github contributions (5)
pytorch/TensorRT
Feb 2021 - Jul 2021
PyTorch/TorchScript/FX compiler for NVIDIA GPUs using TensorRT
Role in this project:
ML Engineer
ML EngineerContributions:7 commits, 13 PRs, 12 pushes in 4 months
Contributions summary:Yu-te primarily contributed to the `pytorch/tensorrt` repository by implementing and testing support for PyTorch operations within the TensorRT environment. This included adding the `aten::transpose` and `aten::layer_norm` operations, enabling more PyTorch models to run efficiently on NVIDIA GPUs. Furthermore, the user addressed linter errors and made general code improvements to ensure code quality and maintainability.
cudapytorchjetsonnvidiadeep-learning
AllentDan/TRTorch
Apr 2021 - Jul 2021
PyTorch/TorchScript compiler for NVIDIA GPUs using TensorRT
Contributions:12 commits in 3 months
cudapytorchnvidiagpucompiler
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.