Rustem Islamov
Looking for autumn/winter internship positions
About me
I am a final-year PhD student at the University of Basel, advised by Aurelien Lucchi. I work at the intersection of the theory and practice of optimization for machine learning. My research focuses on improving our understanding of deep learning optimizers and training techniques, and on making them more efficient and scalable from first principles.
Outside of research, I play tennis and enjoy watching football and judo.
Research Areas
Large-Scale Optimization
See more details →- Scaling Laws from First Principles
- Adaptive Optimizers
- Robustness to Hyperparameter Selection
Loss Landscape
See more details →- Loss Landscape Characterization
- Edge of Stability
Federated Learning
See more details →- Communication Compression and Error Feedback
- Local Methods
Education
Master of Science in Data Science2021 – 2023
Institut Polytechnique de Paris · GPA 17.65/20
Thesis: Unified Analysis of Asynchronous Algorithms
Thesis Supervisors: Mher Safaryan, Dan Alistarh
Bachelor of Science in Applied Mathematics and Physics2017 – 2021
Moscow Institute of Physics and Technology · GPA 4.95/5 (9.27/10)
Thesis: Distributed Second Order Methods with Fast Rates and Compressed Communication
Thesis Supervisor: Peter Richtárik
Recent news
- September 24, 2026I am happy to share that I have two papers accepted to NeurIPS 2026: one on Local SGD and one on Adam-SGD gap.
- August 23, 2026I am attending Swiss Optimization Symposium in Monte Verità . My talk focuses on recent works that use convergence bounds to improve deep learning models training. The slides can be found via the link.
- August 21, 2026I attended UAI 2026 in Amsterdam. I presented my work on Byzantine-robust and differentially private optimization.
- June 29, 2026I am attending ICML 2026 in Seoul. I present four papers during poster sessions on 9th and 10th of July. Moreover, I give an oral presentation on 9th of July on Non-Euclidean Edge of Stability. Finally, I was also invited to give a lighting talk at Protocal Learning workshop.
- June 6, 2026I attended SIOPT 2026. I organized a session on recent advances in federated optimization and gave a presentation on Safe-EF method.
Selected Publications
My contributions have appeared in leading machine learning venues including NeurIPS, ICML, ICLR, and AISTATS. Here is the list of papers I am most proud of: