Chaofan Lin
Research Intern at DeepSeek AI
Shanghai, Shanghai, China
Summary
🤩
Rockstar🎓
Top SchoolChaofan Lin is a PhD student at Tsinghua and a research-focused software engineer with five years' experience building deep learning compilers and systems for LLMs across industry labs including Microsoft Research Asia, ByteDance, and DeepSeek AI. He contributes to Apache TVM as a backend developer and ML engineer, implementing production-facing operators and aligning compiler behavior with PyTorch semantics (e.g., batch_norm training mode and log_softmax/cross-entropy ops). Comfortable moving between research and engineering, he has worked on Relax IR and serving systems for large models, giving him end-to-end insight from compiler kernels to deployment. Based in Shanghai, Chaofan blends rigorous academic training with practical open-source impact, often surfacing subtle correctness and performance fixes that matter in real-world ML stacks.
5 years of coding experience
1 year of employment as a software developer
Doctor of Philosophy - PhD Computer Science, Doctor of Philosophy - PhD Computer Science at Tsinghua UniversityBachelor of Engineering - BE Computer Science (ACM Class), Bachelor of Engineering - BE Computer Science (ACM Class) at Shanghai Jiao Tong University
Github Skills (8)
tvm10
compiler10
machine-learning10
deeplearning-ai10
compiler-compiler10
tensor10
deep-learning10
python10
Programming languages (4)
Github contributions (5)
apache/tvm
Feb 2023 - Dec 2023
Open deep learning compiler stack for cpu, gpu and specialized accelerators
Role in this project:
Back-end Developer & ML Engineer
Back-end Developer & ML EngineerContributions:17 reviews, 20 PRs, 16 comments in 9 months
Contributions summary:Chaofan's contributions center around enhancing the TVM compiler stack with machine-learning related features. They implemented a training mode and momentum argument for the TOPI batch_norm operator, aligning its functionality with torch.nn.functional.batch_norm. Further contributions include the addition of operators such as log_softmax and cross_entropy_with_logits. These changes demonstrate a strong understanding of deep-learning operations within the compiler framework.
metalvulkancompilertensoropencl
SiriusNEO/Masterball
Sep 2021 - Aug 2022
Contributions:182 commits, 166 pushes, 1 branch in 11 months
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.