Sai Karimireddy
Postdoctoral Researcher at University of California, Berkeley
Irvine, California, United States
Summary
🤩
RockstarSai Karimireddy is a postdoctoral researcher at UC Berkeley with a decade of experience applying optimization and machine learning techniques to practical problems. Based in Irvine, California, he blends academic rigor with hands-on engineering, exemplified by implementing Frank-Wolfe algorithm exercises and low-rank matrix completion baselines for an EPFL course repository. His work spans objective design, gradient computation, constrained projections, and projected gradient descent—skills that translate directly to recommendation and large-scale ML systems. Sai’s profile reflects both research depth and practical code contributions, indicating an ability to move theory into reproducible implementations. Colleagues can expect a researcher who writes production-minded experimental code and digs into algorithmic details often glossed over in higher-level research.
10 years of coding experience
Github Skills (4)
machine-learning10
linear-algebra10
python10
gradient-descent10
Programming languages (7)
Github contributions (5)
epfml/OptML_course
Apr 2018 - Apr 2020
EPFL Course - Optimization for Machine Learning - CS-439
Role in this project:
ML Engineer
ML EngineerContributions:28 commits, 23 pushes in 2 years
Contributions summary:Sai implemented practical exercises related to the Frank-Wolfe algorithm in a low-rank matrix completion project. Their contributions involve code modifications to implement several baselines such as global mean, user mean and item mean predictions to solve the movie recommendation using low rank matrix completion problem. Code updates were made to the matrix completion problem where the user needs to compute objective function, the gradient, project onto simplex and trace norm ball, and learn a matrix using projected gradient descent.
optimizationmachine-learningepfl
epfml/error-feedback-SGD
Jan 2019 - Jun 2022
SGD with compressed gradients and error-feedback: https://arxiv.org/abs/1901.09847
Contributions:4 commits, 1 PR, 1 push in 3 years 5 months
arxivabsgradientssgderror-feedback
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.