Yuwei Hu
Deep Learning Compiler Engineer at Edgecortix
United States
Summary
🤩
Rockstar🎓
Top SchoolYuwei Hu is a Deep Learning Compiler Engineer with a decade of experience bridging research and production systems for efficient graph learning and model compilation. Currently at Edgecortix and a PhD candidate at Cornell, Yuwei has deep expertise in optimizing convolution primitives—contributing depthwise convolution and dilation operators to the widely used TVM compiler and implementing numerous operators and mobile examples in NNVM. Past internships at AWS and TuSimple complement a practical focus on making ML models run faster on CPU, GPU, and specialized accelerators. Known for rigorous testing and schedule-level optimizations, Yuwei combines hardware-aware systems thinking with hands-on compiler engineering to squeeze performance from modern deep learning workloads.
10 years of coding experience
Bachelor's degree, Electrical and Electronics Engineering, Bachelor's degree, Electrical and Electronics Engineering at Beihang UniversityDoctor of Philosophy - PhD, Electrical and Computer Engineering, Doctor of Philosophy - PhD, Electrical and Computer Engineering at Cornell University
Github Skills (18)
tvm10
python10
machine-learning10
deeplearning-ai10
compiler-compiler10
deep-learning10
cuda10
compiler10
nvm10
tensor10
convolutional-neural-networks9
performance-monitor9
tensorflow9
neural-network9
performance-analysis9
Programming languages (4)
Github contributions (5)
apache/tvm
Jul 2017 - Sep 2019
Open deep learning compiler stack for cpu, gpu and specialized accelerators
Role in this project:
ML Engineer
ML EngineerContributions:41 commits, 25 PRs, 6 pushes in 2 years 2 months
Contributions summary:Yuwei contributed significantly to the development and testing of depthwise convolution operations within the TVM compiler stack. They implemented and tested depthwise convolution examples, including optimizations and schedule modifications. Their work involved creating test cases and ensuring correctness by comparing results against known implementations, demonstrating a focus on machine learning model optimization and performance tuning within the compiler. They also added dilation operators, which expands the functionalities of the compiler related to convolution operations.
metalvulkancompilertensoropencl
dmlc/nnvm
Sep 2017 - Mar 2018
Role in this project:ML Engineer
ML EngineerContributions:12 commits, 7 PRs, 70 comments in 6 months
Contributions summary:Yuwei primarily contributed to the implementation and testing of various operators within the nnvm framework. Their work focused on registering and integrating new functionalities like softmax, depthwise convolution, elementwise operations, pooling and global pooling. The contributions include writing tests, defining schedules, and ensuring the correct behavior of these operators, demonstrating a strong understanding of deep learning models and compilation techniques. Furthermore, they added a mobile net example using the new APIs of runtime module.
cudametalcomputation-graphtvmdeep-learning
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.