Shigeo Mitsunari

Researcher at Cybozu Labs, Inc.

Japan
email-iconphone-icongithub-logolinkedin-logotwitter-logostackoverflow-logofacebook-logo
Join Prog.AI to see contacts
email-iconphone-icongithub-logolinkedin-logotwitter-logostackoverflow-logofacebook-logo
Join Prog.AI to see contacts

Summary

🤩
Rockstar
🎓
Top School
Shigeo Mitsunari is a researcher and software engineer with 17 years of experience specializing in pairing cryptography, high-performance implementation, and x86/x64 and AArch64 optimization. He is the primary developer of open-source projects such as the JIT assembler Xbyak (used in Intel oneDNN), Xbyak_aarch64 (ported to Fugaku’s A64FX), and the mcl/bls pairing library relied on projects like DFINITY and Ethereum 2.0. His contributions span low-level math and systems work—optimizing oneDNN primitives for AVX512 and aarch64 and integrating xbyak and ate-pairing into libsnark—demonstrating rare cross-domain expertise from cryptographic pairings to CPU-specific numerical tuning. Based in Japan and rooted in a mathematics background from Kyoto University, he blends rigorous theoretical knowledge with practical, production-grade open-source engineering.
code17 years of coding experience
job6 years of employment as a software developer
bookResearch Institute for Mathematical Sciences, Kyoto University
book修士, 数学, 修士, 数学 at 京都大学 / Kyoto University
languagesJapanese
github-logo-circle

Github Skills (16)

neural-network10
avx10
arm10
zksnark10
performance-monitor10
c-language10
performance-analysis10
edn10
cprogramming-language10
deep-neural-networks10
x86-649
x869
build-system9
gnu-make9
api9

Programming languages (13)

C++CRustTeXGoHTMLTypeScriptLLVM

Github contributions (5)

github-logo-circle
uxlfoundation/oneDNN

Apr 2019 - May 2022

oneAPI Deep Neural Network Library (oneDNN)
Role in this project:
userBack-end Developer
Contributions:4 reviews, 13 commits, 1 PR in 3 years 1 month
Contributions summary:Shigeo primarily contributed to optimizing the oneDNN library, focusing on performance improvements for specific CPU architectures. Their work included fixing rounding implementations for the AVX512 instruction set and optimizing the log and ReLU functions for the aarch64 architecture. They also made minor optimizations to the tanh and exp functions.
neural-networkoneapionednndeep-learningdeep-neural-networks
scipr-lab/libsnark

Jul 2014 - Jul 2014

C++ library for zkSNARKs
Role in this project:
userBack-end Developer
Contributions:6 commits, 1 PR, 3 comments in 5 days
Contributions summary:Shigeo contributed to the `libsnark` library, focusing on integration of external libraries and API usage. They integrated the `xbyak` and `ate-pairing` libraries by updating build dependencies and makefiles. Furthermore, they modified core pairing functionality, updating the code to use experimental API calls related to pairing operations in the context of zkSNARKs. These changes indicate a focus on the underlying mathematical operations and their implementation in the library.
cpp
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.
Request Free Trial