Cheng-che Lee
Software Engineer at Meta
Mountain View, California, United States
Summary
🤩
Rockstar🎓
Top SchoolCheng-che Lee is a software engineer with 11 years' experience specializing in distributed training and back-end ML infrastructure, currently on Meta’s MTIA Kernel Team after roles at AWS, Apple, and Amazon. He is a long-standing open-source contributor to Apache MXNet and co-author of the Deep Java Library (DJL), where he implemented data iterator improvements, image preprocessing enhancements, and PyTorch operator support including erfinv. At AWS and Apple he focused on large-scale data-parallel training—contributing to Rufus Pretrain and benchmarking trillion-parameter models with PyTorch FSDP—bringing production-grade distributed training expertise. Based in Mountain View, he combines deep systems-level changes with practical testing and integration work, often addressing edge cases like last-batch handling and rollover behavior in data loaders.
11 years of coding experience
7 years of employment as a software developer
Bachelor's Degree Computer science and information engineering, Bachelor's Degree Computer science and information engineering at National Cheng Kung UniversityMaster’s Degree Software Engineering, Master’s Degree Software Engineering at Carnegie Mellon University
Chinese, English
Github Skills (14)
unit-testing10
javas10
pytorch10
machine-learning10
mxnet10
deep-learning10
python10
java10
ml10
image-processing9
data-science8
tensorflow7
data-structure7
data-structures7
Programming languages (4)
Github contributions (5)
deepjavalibrary/djl
Jun 2019 - Apr 2021
An Engine-Agnostic Deep Learning Framework in Java
Role in this project:
Backend Developer & ML Engineer
Backend Developer & ML EngineerContributions:2 releases, 350 reviews, 406 commits in 1 year 10 months
Contributions summary:Cheng-che's contributions focused on adding missing functions and modifying existing code in the API and MXNet-related directories of the project. This includes creating new functionality to support the DLR (Deep Learning Runtime) engine with code changes related to the image transformations. The user also made enhancements to data pre-processing and included integration tests. Furthermore, the user worked on improvements to the PyTorch-related operators, implemented unit tests for these operations, and added the functionality of the "erfinv" operator.
pytorchmxnetcaffe2deep-learningagnostic
apache/mxnet
Aug 2018 - Aug 2020
Lightweight, Portable, Flexible Distributed/Mobile Deep Learning with Dynamic, Mutation-aware Dataflow Dep Scheduler; for Python, R, Julia, Scala, Go, Javascript and more
Role in this project:Back-end Developer
Back-end DeveloperContributions:3 reviews, 80 commits, 131 PRs in 2 years
Contributions summary:Cheng-che primarily contributed to the `mxnet` repository by modifying the `ImageIter` and `NDArrayIter` classes, which are crucial components for data loading and augmentation. These modifications focused on improving the handling of the last batch, introducing options for "pad," "discard," and "roll_over" behaviors, and addressing potential iteration issues. The commits involved refactoring code, adding unit tests, and enhancing the functionality of data iterators. The user also worked on the SN-GAN example by adding comments and improving the example.
pythonschedulerdataflowmutationdata-science
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.