Wei-shengĀ Chin
Principal Machine Learning Scientist at Microsoft
Redmond, Washington, United States
Summary
š¤©
Rockstarš
Top SchoolWei-sheng Chin is a Principal Machine Learning Scientist with 11 years of experience, currently based in Redmond and focused on accelerating LLM inference and training pipelines at Microsoft. He blends deep research training (PhD and postdoc) with pragmatic engineering, shipping performance and interoperability improvements across high-profile open-source projects like ONNX, PyTorch and ONNX Runtime. His contributions span operators, exporters, optimizer correctness (Adam/Lamb bias correction) and model conversion tooling, reflecting a rare combination of low-level backend systems and ML algorithm expertise. Known for fixing subtle shape-inference and graph-export bugs, he helps bridge research models to production runtimes at scale.
10 years of coding experience
6 years of employment as a software developer
Doctor of Philosophy (PhD), Computer Science, Doctor of Philosophy (PhD), Computer Science at National Taiwan UniversityMaster of Science (M.S.), Electrical and Electronics Engineering, Master of Science (M.S.), Electrical and Electronics Engineering at National Tsing Hua University
English, Chinese
Github Skills (28)
pytorch10
c-language10
python10
scikit10
machine-learning10
inference10
onnx10
netframework10
deep-learning10
compile10
lightgbm10
dotnet-core10
scikit-learn10
compiler10
model-conversion10
Programming languages (8)
Github contributions (5)
dotnet/machinelearning
Oct 2018 - Aug 2019
ML.NET is an open source and cross-platform machine learning framework for .NET.
Role in this project:
ML Engineer
ML EngineerContributions:124 commits, 240 PRs, 106 pushes in 10 months
Contributions summary:Wei-sheng's primary contributions revolve around enhancing the functionality of the ML.NET framework, specifically concerning the creation and use of ONNX initializers within the `dotnet/machinelearning` repository. The user implemented the creation of ONNX initializers, enabling the creation of constant tensors within the graph, and integrated these into the model definition. Additionally, the user introduced a test case, demonstrating the creation of various initializers and their proper use, showcasing the integration of the new functionality.
machine-learning-platformdotnetml-netmachine-learningcsharp
onnx/onnx
Apr 2018 - Jun 2022
Open standard for machine learning interoperability
Role in this project:ML Engineer
ML EngineerContributions:77 reviews, 34 commits, 111 PRs in 4 years 2 months
Contributions summary:Wei-sheng contributed to the ONNX project by implementing and modifying machine learning operators. Their work includes adding an IsInf operator for detecting infinity values, upgrading the LabelEncoder operator to support more input types, and fixing shape inference issues. They also added bit-shift operators, reflecting a focus on features supporting more sophisticated machine learning functionalities.
pytorchmxnetdeep-learninginteroperabilitymachine-learning
Find and Hire Top DevelopersWeāve 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.