Michael Shvartsman
Research Scientist, FAIR at Meta
Seattle, Washington, United States
Summary
👤
Senior🎓
Top SchoolMichael Shvartsman is a Research Scientist at Meta FAIR with 11 years of cross-disciplinary experience bridging theoretical ML, neuroscience, and product-facing research. He currently leads the human data program for FAIR’s AI Research Agents project, combining rigorous empirical work with tool-building to make agent research reproducible and safe. His background—PhD-level training in psychology, postdoctoral neuroscience research at Princeton, and roles as a research lead and manager at Meta—gives him a rare ability to translate cognitive science insights into practical ML systems. An active open-source contributor, he’s improved numerical robustness in gpytorch and modernized pymanopt for TensorFlow 2 eager mode, reflecting deep care for reliable scientific software. Based in Seattle, he blends theory and implementation to move foundational ideas into production-ready tooling.
11 years of coding experience
11 years of employment as a software developer
Doctor of Philosophy (PhD), Doctor of Philosophy (PhD) at University of MichiganB.A., B.A. at Yale University
Github Skills (21)
autodiff10
pytorch10
python10
back-end-development10
algebra10
gaussian-processes10
numerics10
tensorflow10
numerical-methods10
numerical-analysis10
numerical-computing10
optimization10
decomposition9
deco9
matrix-decomposition9
Programming languages (6)
Github contributions (5)
cornellius-gp/gpytorch
May 2019 - May 2021
A highly efficient implementation of Gaussian Processes in PyTorch
Role in this project:
ML Engineer
ML EngineerContributions:2 reviews, 7 commits, 4 PRs in 1 year 11 months
Contributions summary:Michael's contributions primarily involve modifying and refining core components of the Gaussian Processes implementation in PyTorch. They addressed issues related to numerical stability and accuracy in the `psd_safe_cholesky` function and related utilities. Additionally, the user worked on the `IndexKernel`, ensuring its proper functioning in non-square cases and fixing related bugs. These changes indicate a focus on improving the robustness and functionality of the core library.
pytorchgpu-accelerationgaussianstochastic-processesmodular
pymanopt/pymanopt
Aug 2020 - Oct 2020
Python toolbox for optimization on Riemannian manifolds with support for automatic differentiation
Role in this project:ML Engineer
ML EngineerContributions:5 commits, 1 PR, 6 comments in 1 month
Contributions summary:Michael's primary contribution involved updating the TensorFlow backend for the pymanopt library to support TensorFlow 2 eager mode. They modified the TensorFlow backend code, updated tests, and incorporated the new TensorFlow backend into various examples within the repository. This work included adapting existing code to the new eager execution model and ensuring compatibility.
pythonautomatic-differentiationdifferentiationtoolboxoptimization
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.