Grigory Malinovsky
Bio
I am a postdoctoral researcher at EPFL, working with Prof. Volkan Cevher in LIONS on optimization for machine learning. I received my PhD in Applied Mathematics and Computational Sciences from King Abdullah University of Science and Technology (KAUST) in June 2026, advised by Prof. Peter Richtárik.
I develop optimization methods for efficient and reliable machine learning. My work connects stochastic and adaptive methods with communication-efficient federated learning, privacy guarantees, and robustness to unreliable participants.
During my PhD, I worked as a Research Intern Engineer at Samsung R&D Institute UK from April to October 2024, focusing on optimization theory for large language models.
My hobbies include fitness, basketball, drums, and piano.
Recent news
- — Joined EPFL as a postdoctoral researcher in LIONS, working with Volkan Cevher.
- — New preprint: How Does Muon Work in Private Deep Learning?
- — Completed my PhD at KAUST, advised by Peter Richtárik.
Education
PhD in Applied Mathematics and Computational Sciences
Computer, Electrical and Mathematical Sciences and Engineering Division
King Abdullah University of Science and Technology
Jan. 2022 – Jun. 2026
Thuwal, Saudi Arabia
Thesis: “Theoretical Foundations of Communication-Efficient, Robust, and Practical Distributed and Federated Optimization”
Advisor: Peter Richtárik
MS in Applied Mathematics and Computational Sciences
Computer, Electrical and Mathematical Sciences and Engineering Division
King Abdullah University of Science and Technology
Aug. 2020 – Dec. 2021
Thuwal, Saudi Arabia
GPA: 3.83/4.0
Advisor: Peter Richtárik
BS in Applied Mathematics and Physics
Department of Control and Applied Mathematics
Moscow Institute of Physics and Technology
Sept. 2014 – Jul. 2019
Dolgoprudny, Russia
GPA: 4.73/5.0
Thesis: “Averaged Heavy Ball Method”
Advisor: Boris Polyak
Academic Service
- Reviewing: ICML, NeurIPS, ICLR, AISTATS, Transactions on Machine Learning Research, and IEEE Transactions on Automatic Control.
- Program committees: IEEE FLTA 2024 and federated learning workshops at ICML 2023 and NeurIPS 2022.
- Seminar organization: OPT ML group seminar, 2022.