Hongbin Zhang
Postdoctoral Researcher at 中国科学院软件研究所
Beijing, China
Summary
🤩
Rockstar🎓
Top SchoolHongbin Zhang is a postdoctoral researcher and leader of the Ruyi AI Infrastructure / PLCT MLIR team at the Institute of Software, Chinese Academy of Sciences, with nine years of experience building compiler and AI infrastructure. He specializes in MLIR-based compiler engineering and backend optimizations, notably implementing a convolution optimization pass using a Coefficients Broadcasting with Strip Mining (CB-SM) technique in the buddy-mlir framework. Beyond hands-on compiler work, he plays an influential role in open hardware and standards as a member of RISC-V International's TSC and a CHIPS Alliance governing board member. His background blends rigorous academic training (PhD in Software Engineering) with practical systems experience from internships at Red Hat and international study in the U.S. Based in Beijing, he bridges research and production by shipping compiler passes that target domain-specific architectures. Colleagues describe him as a pragmatic engineer who turns theoretical optimizations into reproducible toolchain improvements.
9 years of coding experience
2 years of employment as a software developer
Ph.D. Software Engineering, Ph.D. Software Engineering at University of Chinese Academy of SciencesOther Summer School, Other Summer School at San José State UniversityBachelor's Degree Computer science, Bachelor's Degree Computer science at Beijing University of Technology
Github Skills (6)
compiler-optimization10
convolution10
c-language10
cprogramming-language10
vectorization10
mlr10
Programming languages (11)
Github contributions (5)
buddy-compiler/buddy-mlir
Apr 2021 - Dec 2022
An MLIR-based compiler framework bridges DSLs (domain-specific languages) to DSAs (domain-specific architectures).
Role in this project:
Back-end Developer & Compiler Engineer
Back-end Developer & Compiler EngineerContributions:261 reviews, 615 commits, 180 PRs in 1 year 7 months
Contributions summary:Hongbin implemented a convolution optimization tool within the buddy-mlir compiler framework, leveraging the Coefficients Broadcasting with Strip Mining (CB-SM) approach. This involved introducing a new pass for vectorization, modifying the compilation driver, and adding example code. The work demonstrates a focus on optimizing convolution operations within the MLIR framework. The contributions include adding a new pass and modifying existing code.
mlir
buddy-compiler/buddy-benchmark
Sep 2021 - Dec 2022
Benchmark Framework for Buddy Projects
Contributions:75 reviews, 70 commits, 43 PRs in 1 year 3 months
benchmarkbuddybenchmark-framework
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.