Lixun Zhang
Member Of Technical Staff at AMD
Austin, Texas, United States
Summary
🤩
Rockstar🎓
Top SchoolLixun Zhang is a machine learning compiler engineer with a decade of experience bridging GPU architecture, compiler backend work, and performance modeling, currently a Member of Technical Staff at AMD in Austin. His work on the Triton compiler and AMD GPU code generation demonstrates deep expertise in instruction selection, MFMA optimizations, denorm handling, and epilogue passes to squeeze real-world performance from hardware. He pairs academic rigor—PhD in Computer Science from UT Austin and research on particle filters, analytical GPGPU models, and heterogeneous implementations—with hands-on C++ compiler engineering from internships at MathWorks to production contributions. Comfortable moving between research and production, Lixun has a track record of automating translation tools and practical performance fixes that reveal subtle hardware-software interactions.
10 years of coding experience
9 years of employment as a software developer
Doctor of Computer Science, Computer Science, Doctor of Computer Science, Computer Science at The University of Texas at AustinBachelor of Engineering (BEng), Automation Engineer Technology/Technician, Bachelor of Engineering (BEng), Automation Engineer Technology/Technician at Tsinghua University
English, Chinese
Github Skills (9)
compiler-optimization10
triton10
amdgpu10
llvm9
backend9
back-end-development9
performance-tuning9
roc8
hip7
Programming languages (7)
Github contributions (5)
triton-lang/triton
May 2023 - Mar 2025
Development repository for the Triton language and compiler
Role in this project:
Back-end Developer & Performance Engineer
Back-end Developer & Performance EngineerContributions:392 reviews, 133 PRs, 130 pushes in 1 year 10 months
Contributions summary:Lixun contributed to the Triton language and compiler by modifying and refactoring backend code. They focused on optimizing the compiler and improving performance through instruction selection and code generation for AMD GPUs. Key contributions include refactoring MFMA selection logic, improving denorm handling, and optimizing the epilogue pass. Furthermore, they addressed and resolved issues in various components, including the scanOp lowering process and handling cache modifiers for tl.load.
compilerprogramming-languagecode-generationtriton
zhanglx13/triton
Aug 2023 - Mar 2025
Development repository for the Triton language and compiler
Contributions:152 pushes, 46 branches in 1 year 7 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.