Junior Data Scientist (математик)

Описание

Minau Yauheni
+375296340781; minov.03012000@gmail.com ; telegram : @Evg_M
M achine Learning | Python | C++ | SQL | Amazon Web Services | Git
Work Experience
M achine Learning Engineer (EPAM Systems ). March 2022 – March 2023
Projec t: text classification for streamlin ing people -related processes .
1. Revisited the business problem formulation together with the BA .
Chose the metric and validation scheme to match the business scenarios (including time -series cross -
validation).
2. Reviewed the upstream entity extraction model s, found and reported several issues , worked together with
the model authors on the fixes.
3. Optimized the pipeline to reduce train ing time and streamline experiments (including hyperparameter
tuning):
 Profiled the pipeline to identify bottlenecks .
 Sped up model fitting process 200 0 times by using different solvers, libraries (Snap ML, Intel®
Extension for Scikit -learn), sparse data formats and data preparation tweaks (scaler and non -linear
transformation) .
 Sped up data preparation 50 times by implementing own data subsetting methods and changing the data
format .
The overall running time went down 12-20 times compared to the original implementation.
4. Conducted an extensive set of experiments for choosing preprocessing options, specific models and their
hyperparameters using Optuna and MLFlow. Improved the target metric by 5% (despite the existing
consensus that the maximum had been reached and no improvements could’ve been obtained even by
applying very resource -intensive approaches like deep neural networks ).
5. Automated part of the entity extraction model retraining pipeline.
Technologies: Python, Scikit -Learn, pandas, N um Py, Sci Py, O ptuna, M atplotlib, seaborn,
plotly, MLflow, PostgreSQL, NLTK , Spacy, AWS, Apache A irflow, Jira .
Research Engineer (Huawei ). July 202 1 – February 202 2
Project: development and optimization of the Mixed -Integer Programming solver.
1. Designed data structures for implementing a new complex algorithm to speed up a Mixed -Integer
Programming solver.
2. Implemented the Conflict Analysis algorithm in the MIP solver .
3. Researched performance optimization approaches .
Technologies: C++, Clion, Xftp, Xshell.
Software Engineer (Huawei ). November 202 0 – June 202 1
Project: offline maps application development .
1. Adapted the format of the available geo -data for the A-star path search algorithm .
2. Tested correctness of the correspondence of geo structures to their real objects.
3. Tested the A-star algorithm on the available data. Found and reported several issues.
4. Ported the module from Java to C++ to improve performance.
Technologies: C++, Java, Python, G eo Py.

Education
Belarusian State University of Informatics and Radioelectronics ,
Department of Computer Systems and Networks .
Bachelor’s degree, GPA math 10.0 / 10.0; GPA others 7. 7 / 10 .
- C++, Python, Java, SQL .
- Calculus, Algebra and Geometry, Analysis , Probability theory and Statistics.
A wards
 International Mathematics Competition for University Students: 1st degree (2022, 2020, 2019); 2nd
degree (2021) .
 Republican Belarusian Mathematical Olympiad for students: 1st absolute place (2019) .
 BSUIR Mathematical Olympiad for students: 5 absolute first places (2018 -2021) .
Teaching experience
Train ed students for Olympiads in higher mathematics .
Topics at lessons: linear algebra, limits and series, functional analysis, complex analysis, probability theory,
combinatorics, graph theory and group theory. Information related with these topics have olympiad difficulty
level.
 At 2019 (H1) train ed studen ts of BSUIR .
Obtaine d bronze medal at IMC 2019.
 At 2019 (H2) train ed students of BSUIR and BSU .
Obtain ed two honorable medals at IMC 2020.
 At 2021 (H2) and 2022 (H1) train ed remotely students of BSUIR, BSU and KNU .
Obtained three 1st degree diplomas and one 2nd degree diploma at IMC 2022 .
 At 2023 (H1) train remotely students of BSUIR, BSU and KNU.