zianglih (Ziang Li)
github.com · 246 words · saved by 1 readers
@HumansAnd | ex-@NVIDIA DevTech | ex-@google-gemini | @cuhkshenzhen - zianglih
Hi there 👋 I am 栗子昂, an MTS @ humans& ai working on the full-stack of LLM performance engineering. I graduated from University of Michigan, Ann Arbor (BS), after I transfered and spent 2 years at Chinese University of Hong Kong, Shenzhen. [News] Check out my invited talk at GPU MODE: https://youtu.be/wiaUh82NEoE?si=gVBqCqZm092l1bbA [News] Check out our NVFP4 RL recipe blogs: https://humansand.ai/blog/nvfp4-rl https://www.lmsys.org/blog/2026-07-29-mxfp8-nvfp4-rl Something I have been working on recently: nvfp4 RL recipe mxfp8 RL topk for sparse attention Some of the places I…
saved by
related reading
- Xianbang Wang — research & writingpeppaking8.github.io
- Lil'Loglilianweng.github.io
- Vincent Huangvvhuang.com
- Horace He: Building Machine Learning Systems for a Trillion Trillion Floating Point Operationsyoutube.com
- ⭐️ Fast LLM Inference From Scratchandrewkchan.dev
- Chip Huyenhuyenchip.com
- Machine Learning System Resources | std::bodun::blogbodunhu.com
- LYDIAlydia.ml
- i care aboutbuilding things that matter.yiannisha.dev
- Chunyuan Lichunyuan.li
- Nicholas Carlininicholas.carlini.com
- GitHub - linkedin/Liger-Kernel: Efficient Triton Kernels for LLM Traininggithub.com