Shintaro Iwasaki
Research Scientist at Meta
San Francisco Bay Area United States
Summary
🤩
Rockstar🎓
Top SchoolShintaro Iwasaki is a research scientist based in the San Francisco Bay Area with a decade of experience building high-performance systems for scientific computing and machine learning. He brings deep expertise in threading, synchronization, and low-level performance engineering—evidenced by substantive contributions to flagship MPI implementations (Open MPI, MPICH) and CPU/GPU kernel optimizations in FBGEMM and Triton. His work spans production-grade package management and reproducible builds via Spack to cutting-edge ML compiler/runtime improvements, reflecting a rare blend of systems-level rigor and ML performance tuning. A University of Tokyo PhD, he transitioned from sustained research roles at Argonne National Laboratory to industry research at Meta, combining academic depth with practical impact on widely used open-source projects. An often-overlooked strength is his focus on portability and tooling—removing dependencies and improving build/configuration hygiene to make complex stacks more maintainable and performant.
10 years of coding experience
University of Tokyo
Japanese, English
Github Skills (45)
pytorch10
c-language10
system-programming10
python10
multithreading10
atomics10
fortran10
gpu-programming10
mpi10
machine-learning10
hpc10
c1110
c1710
hpcc10
package-manager-tool10
Programming languages (9)
Github contributions (5)
pytorch/FBGEMM
Sep 2021 - Jan 2023
FB (Facebook) + GEMM (General Matrix-Matrix Multiplication) - https://code.fb.com/ml-applications/fbgemm/
Role in this project:
ML Engineer
ML EngineerContributions:62 reviews, 31 commits, 49 PRs in 1 year 4 months
Contributions summary:Shintaro primarily focused on optimizing and extending the functionality of the fbgemm library, which includes FB (Facebook) + GEMM (General Matrix-Matrix Multiplication). Their contributions involved vectorizing and optimizing kernel functions for CPU and GPU, resulting in performance improvements. They addressed issues related to sparse matrix operations and benchmarking tools to evaluate the performance of quantized operations. Additionally, they enhanced the library by open-sourcing new features such as `segment_sum_csr()` and improved code generation for optimized kernels.
matrix-multiplicationfacebookmultiplicationmatrixml-applications
pmodels/mpich
Feb 2019 - Jul 2021
Official MPICH Repository
Role in this project:Back-end Developer
Back-end DeveloperContributions:114 reviews, 285 commits, 100 PRs in 2 years 5 months
Contributions summary:Ken Raffenetti primarily worked on improving the MPICH library's threading and atomic operations. He addressed threading initialization issues by adding flags and interfaces for thread initialization and finalization, particularly for libraries like Argobots. His contributions also included the creation of a new MPL atomic wrapper, which is aimed at deprecating OpenPA and provides support for various atomic operations. Furthermore, Ken integrated these new thread initialization and finalization interfaces within MPI_Init and MPI_Finalize.
fortranhpcmpic
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.