CV
Education
- PhD (Computer Science): Constructor University, Bremen, Germany, 2024-2027 (expected)
- MSc (Math of Machine Learning): HSE University, Moscow, Russia, 2021-2023
Thesis title: Ensembling Neural Networks in the Transfer Learning Setup - BSc (Applied Mathematics & Informatics): HSE University, Moscow, Russia, 2017-2021
Thesis title: Empirical Analysis of Self-Supervised Training Properties
Languages: Russian (native), English (C1), German (actively learning, ~B1)
Work experience
- Centre of Deep Learning and Bayesian Methods — HSE University, Moscow, Russia, 2021-2023
Research assistant: Fundamental research in deep learning, loss landscapes, training dynamics - SmartDec — Moscow, Russia, 2019-2020
Intern, development & analytics: Industrial research in cryptography and zero-knowledge proofs
Skills
- Programming languages:
- Proficient: Python, LaTeX
- Intermediate: C/C++, Rust
- Technologies:
- Data science: PyTorch, NumPy, Pandas, scikit-learn, matplotlib
- Linux: bash, git, ssh, docker
Publications
Talks
Can Stationary Distributions of Scale-Invariant Neural Networks Be Described by the Thermodynamics of an Ideal Gas?
Paper presentation at 35th International Joint Conference on Artificial Intelligence, Bremen, Germany
To Stay or Not to Stay in the Pre-train Basin: Insights on Ensembling in Transfer Learning
Paper presentation at Fall into ML 2023: Conference on Machine Learning at HSE University, Moscow, Russia
Self-supervised Pre-training with Masked Image Modeling
Research seminar at Bayesgroup, online
On the Memorization Properties of Contrastive Learning
Poster session at Workshop on Overparameterization: Pitfalls & Opportunities at ICML, online