Amir Saadat
Software Engineer at Google
San Francisco Bay Area United States
Summary
👤
Senior🎓
Top SchoolAmir Saadat is a software engineer and computational scientist with a decade of experience bridging high-performance computing, numerical methods, and machine learning. He combines a deep academic background (5+ years PhD and ~3.5 years postdoc at Stanford) with four years in industry, currently building core ML and XLA infrastructure at Google. His work spans GPU/TPU-accelerated solvers, parallelized C++/CUDA/Fortran codes, and Scientific ML/Bayesian models applied to biophysics and computational biology. Notably, he contributed advanced optimizers—Gauss-Newton and Levenberg-Marquardt solvers—to the widely used google/jaxopt project, underscoring expertise in hardware-accelerated numerical optimization. Colleagues rely on him to translate complex multi-physics problems into production-grade, high-performance software that runs at scale.
10 years of coding experience
8 years of employment as a software developer
Ph.D. M. Sc. Chemical and Biomolecular Engineering, Ph.D. M. Sc. Chemical and Biomolecular Engineering at University of Tennessee, KnoxvilleAmirkabir University of TechnologyPostdoctoral Scholar Chemical Engineering, Postdoctoral Scholar Chemical Engineering at Stanford University
Spanish, English, German
Github Skills (10)
algorithm10
differentiable-programming10
jax10
python10
optimisation10
optimizers10
optimization10
linear-algebra9
numerical-methods9
deep-learning8
Programming languages (1)
Github contributions (5)
google/jaxopt
Jan 2022 - Oct 2022
Hardware accelerated, batchable and differentiable optimizers in JAX.
Role in this project:
ML Engineer
ML EngineerContributions:53 reviews, 7 commits, 8 PRs in 9 months
Contributions summary:Amir implemented and refined optimization algorithms within the JAX framework. Their contributions included the creation of a Gauss-Newton least-squares solver and a Levenberg-Marquardt solver, demonstrating expertise in numerical optimization techniques. The commits involve significant code changes within the project's core, adding new functionality to the library. The work clearly aligns with the project's goal of providing hardware-accelerated optimizers.
pytorchdifferentiabledifferentiable-programmingautomatic-differentiationdeep-learning
amir-saadat/jaxopt
Jan 2022 - Aug 2023
Hardware accelerated, batchable and differentiable optimizers in JAX.
Contributions:60 pushes, 7 branches in 1 year 7 months
pytorchdifferentiableoptimizationacceleratedmachine-learning
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.