cheng cheng
Software Engineer at Skywork
Beijing, China
Summary
🤩
Rockstarcheng cheng is a software engineer based in Beijing with 11 years of hands-on experience building performant back-end systems. A Tsinghua-trained "master of software," he contributes to large-scale open-source deep learning infrastructure, notably implementing core memory management and compute kernels for the OneFlow framework. His work on TensorBuffer and optimized operators like ReLU demonstrates a focus on low-level performance, efficiency, and maintainability in ML runtimes. Comfortable navigating complex system design, he blends algorithmic thinking with pragmatic engineering to push scalable features into production. Colleagues would describe him as an engine-level problem solver who prefers refining foundational components over surface-level polish.
10 years of coding experience
Github Skills (8)
kernel10
cuda10
memory-management10
machine-learning10
c-language10
deep-learning10
cprogramming-language10
numerical-computing10
Programming languages (5)
Github contributions (5)
Oneflow-Inc/oneflow
Apr 2017 - Jan 2023
OneFlow is a deep learning framework designed to be user-friendly, scalable and efficient.
Role in this project:
Back-end Developer
Back-end DeveloperContributions:1 release, 1411 reviews, 1068 commits in 5 years 9 months
Contributions summary:Cheng primarily contributed to the OneFlow deep learning framework by implementing and refactoring several core components related to memory management and computation within the system. They were involved in the design and implementation of new features like the TensorBuffer data type and its related operations. Their work included implementing mathematical operations and kernels, such as those for the ReLU, and refining existing functionality by optimizing for performance and efficiency.
caffe2tensorflowframework-learningchainerdeep-learning-framework
SkyworkAI/Skywork
Oct 2023 - Mar 2025
Skywork series models are pre-trained on 3.2TB of high-quality multilingual (mainly Chinese and English) and code data. We have open-sourced the model, training data, evaluation data, evaluation methods, etc. 天工系列模型在3.2TB高质量多语言和代码数据上进行预训练。我们开源了模型参数，训练数据，评估数据，评估方法。
Contributions:4 reviews, 6 PRs, 9 pushes in 1 year 4 months
llm
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.