Lianmin Zheng
Research Scientist at Meta
San Francisco Bay Area United States
Summary
🤩
Rockstar🎓
Top SchoolLianmin Zheng is a research scientist in the Bay Area with a decade of experience building high-performance ML systems and developer-facing infrastructure. He has driven inference and pretraining infrastructure at xAI (including leadership on a 100k GPU pretraining stack) and now applies research at Meta’s MSL TBD Lab, combining deep systems engineering with practical ML research. His open-source contributions span TVM compiler optimizations, distributed training (Alpa), and serving frameworks for large language and vision models, reflecting a focus on performance, compiler/runtime work, and developer efficiency. Notably, he has fixed subtle model-serving bugs and improved OpenAI-compatible chat handling in SGLang, showing attention to UX in low-level ML tooling. He holds a PhD candidate background at UC Berkeley and brings a blend of academic rigor and production-grade engineering to scale ML workloads.
10 years of coding experience
4 years of employment as a software developer
Doctor of Philosophy - PhD, Computer Science, Doctor of Philosophy - PhD, Computer Science at University of California, BerkeleyBachelor of Science - BS, Computer Science, Bachelor of Science - BS, Computer Science at Shanghai Jiao Tong University
Chinese, English
Github Skills (35)
tvm10
pytorch10
c-language10
performance-monitor10
performance-analytics10
back-end-development10
machine-learning-models10
python10
mxnet10
machine-learning10
distributed-systems10
reinforcement-learning10
performance-measurement10
code-generation10
performance-analysis10
Programming languages (13)
Github contributions (5)
alpa-projects/alpa
Feb 2021 - Jan 2023
Training and serving large-scale neural networks with auto parallelization.
Role in this project:
ML Engineer
ML EngineerContributions:11 releases, 195 reviews, 295 commits in 1 year 11 months
Contributions summary:Lianmin's commits primarily focus on the development and testing of distributed machine learning models within the context of the Alpa project. They introduced and tested several new features to the framework including support for various models such as the Conformer and further support for the existing Bloom model. Their work involved improvements to the core functionalities of automatic sharding and gradient accumulation.
alpatputrainingcompilertensorflow
sgl-project/sglang
Jan 2024 - Apr 2025
SGLang is a fast serving framework for large language models and vision language models.
Role in this project:Back-end Developer & ML Engineer
Back-end Developer & ML EngineerContributions:6 releases, 709 reviews, 1705 PRs in 1 year 2 months
Contributions summary:Lianmin made several significant contributions to the SGLang project, including bug fixes related to the handling of logprobs within the code. The user also made improvements to error messages to enhance the user experience. The user implemented code changes and improvements related to supporting OpenAI, particularly chat model handling, and provided documentation.
cudadeepseekdeepseek-llmdeepseek-r1deepseek-r1-zero
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.