About Me
Hi, I am PhD student and research scientist at Moscow Institute of Physics and Technology. Currently I'm working as machine learning engineer at Huawei Russian Research Instituse. My academia research is based on classical machine learning approaches: feature selection, object recognition, model verification, computational complexity. Last year I finished Yandex data school.
Experience
Intelligent systems and Data science Technology center - Computing algorithm team
Our project is related to the development and optimization of distributed algorithms for the Apache Spark framework. I'm involved into subproject: development and optimization of distributed classical machine learning. We develop a general-purpose library (collection) of algorithms that will be used by Huawei customers to analyze their real-world data
MIPT, one of the leading Russian universities in the areas of physics and technology. The university holds a reputable position in the country and abroad in qualified graduate training.
There are several tasks, that I was involved
- Creation of a unified methodology of constructing algorithms for recovering a wide class of dependencies in terms of generalized linear models.
- Development of a methodology for a differential cross-validation scheme for the selection of hyperparameters.
- Implementation of the scheme for hyperparameters estimation by differential cross-validation.
Math tutor
Undergraduate Research Intern
October 2013 - May 2018
I worked with many pupils from 1st to 11th. We trained logical thinking, both basic and advanced math. Also taught olympiad mathematics, after that some of my students joined MIPT.
- Teacher of math and physics at physics and technology school.
- Reviewer of Lomonosov olympiad (math) in Moscow.
- Jury member, organisator, reviewer of Ukraine olympiad movement in Kiev.
ZNOUA - is an educational company with a great mission: to give every child in Ukraine the opportunity to enter the school of their dreams, thus changing their future for the better!
There were three classes of students: two on base and one in advanced track. All of them successfully passed exams end became university students.
Education
Working on PhD thesis: "Verification of non-stationary dependency models". Got several publications and conference presentations, more details below.
With Master's degree I acquired knowledge on data science and machine learning. Ended with diploma "Algorithms for choosing a regression model in big data" and began studying on PhD.
At university I've got classical computer science education at faculty of computer science and cybernetics. After second cource I joined Department of Computational Mathematics and made my diploma "Research of random number sensors".
Additional education
Joined YSDA to learn most advanced and cutting edge knowledge on both classical machine learning and deep learning.
The training program in the basics of risk management in the financial sector that covers all major areas of risk management. As part of this, I received applied knowledge in the field of risk management from the best experts of Sberbank.
Passed two courses:
- Risk management. Studied forecasting customer behavior, predicting the solvency of borrowers, evaluating the profitability of new business areas.
- Fintech trends. Lectures about how big tech companies create services and applications from top-managers Tinkoff.
Research
Main publications
- Constrained Regularized Regression Model Search in Large Sets of Regressors
- Sparse constrained regression; Returns based analysis of invest portfolios; Investment portfoliocomposition
- DOI: 10.1007/978-3-319-96133-0_30
- Springer
- July 2018
- Computational Complexity of Dependence Estimation via Generalized Linear Models in Multidimensional Feature Spaces
- computational complexity, machine learning, pattern recognition, regression analysis
- DOI: 10.1109/SIBIRCON48586.2019.8958417
- IEEE
- October 2019
- Differential Leave-One-Out Cross-Validation for Feature Selection in Generalized Linear Dependence Models
- dependence estimation, model verification, feature selection, LOO-CV
- DOI: 10.1145/3473465.3473474
- ITCC (Information Technology and Computer Communications)
- October 2021
Conferences
- Algorithmic Implementation of Factor Search in Returns Based Analysis of Investment Portfolios
- Intelligent Data Processing: Theory and Applications. 12th International Conference
- Gaeta, Italy, October 2018
- DOI: 10.30826/IDP201847
- Slides
- Factor Search in Returns Based Analysis of Investment Portfolios
- Intelligent Data Processing: Theory and Applications. 12th International Conference
- Gaeta, Italy, October 2018
- DOI: 10.30826/IDP201845
- Slides
- On-line estimation of generalized linear dependence models from growing training sets
- Mathematical Methods for Pattern Recognition. 19th Russian National Conference with International Participation
- Moscow, Russia, November 2019
- Slides
- Method of differential leave-one-out cross validation for chosing the complexity level in generalized linear models of dependences
- Mathematical Methods for Pattern Recognition. 19th Russian National Conference with International Participation
- Moscow, Russia, November 2019
- Slides
Other skills and interests
Technical skills
- Programming: Python Scala C++ C Latex
- Libraries: Spark pytorch keras transformers plotly neptune ampligraph dask jax skimage nltk ...
Languages
- Russian: Native
- Ukrainian: Native
- English: Advanced
Grants (RFBR)
- 19-37-90159: "Linear complexity algorithms for dependency estimation from large amount of empirical data"
- 20-07-00382: "Methods for reducing the computational complexity of dependency estimation algorithms in big empirical data"