Aleksei Rodchenko

Регистрация: 26.03.2022

Специализация: Data Scientist

Портфолио

Sberbank AI-dashboard

Developed time-series model for forecasting Sberbank's business indicators. Created Oracle database scheme. Developed a backend of an intelligent dashboard for NRT detecting abnormal behavior of 20,000 bank's business indicators using forecast data. Dashboard shows only indicators being currently abnormal.

SDCM

Created a ground uplink station (GUS) for transmitting SDCM information (SDCM is a system for improvement of GPS and GLONASS accuracy, reliability and availability on the territory of Russia) to communication satellites (hardware and software development). Performed modeling and preliminary design of SDCM satellite channels (satellite link budget calculation, satellite service area modelling). Managed the development of a second generation SDCM resulting in decrease the error of satellite navigation from 1.5 m to 0.5 m, and decrease the time of integrity fail warning from 10 to 6 seconds (according to first generation SDCM)

Yandex Praktikum projects

Some of the projects made during Yandex training course. EDA, regression tasks, classification tasks, time-series, NLP, computer-vision.

Скиллы

C++

CatBoost

Confluence

Data Scientists

Docker

fbProphet

Git

Jenkins

Jira

Keras

LightGBM

Linux shell

Matlab

Matplotlib

Numpy

OpenCV

Optuna

Oracle SQL

Pandas

Plotly

PostgresSQL

Python

PyTorch

RestAPI

Scipy

Seaborn

Sklearn

SQL

Statsmodels

Swarm

Tensorflow

WebSocket

Опыт работы

Data Scientist

с 04.2021 - По настоящий момент |Sberbank of Russia

Python, Jupyter Notebook, LightGBM, fbProphet, docker, Jenkins, Jira, Bitbucket, Confluence, Oracle, Postgres

- Time series forecasting, incidents detecting, python backend development, database engineering. - Developed a fast and universal time-series model for forecasting Sberbank's business indicators. This model increased the amount of forecasted indicators from 500 to 20,000. - Developed a backend of an intelligent dashboard for detecting abnormal behavior of the bank's business indicators using forecast data. Dashboard shows only indicators being currently abnormal (usually up to 10 simultaneously of 20,000 total) and therefore helps operators to reduce manual incidents detection time. - Improved a model for automatic detection of the bank incidents using multiple business indicators. The improved model reduced the average incident detection time by 2 minutes. - Developed a pipeline for collecting business indicators and infrastructure indicators (CPU load, network interfaces load etc.) and network configuration to use in Sberbank’s infrastructure Digital Twin.

Head of Hardware and Software Development Department

10.2014 - 12.2020 |Russian Space Systems

SDCM, GLONASS, GPS

- Managed the development of a second generation SDCM (SDCM is the system for improvement of GPS and GLONASS accuracy, reliability and availability on the territory of Russia, SDCM - Navipedia) resulting in decrease the error of satellite navigation from 1.5 m to 0.5 m, and decrease the time of integrity fail warning from 10 to 6 seconds (according to first generation SDCM). - Performed modeling (MATLAB) and preliminary design of third generation SDCM satellite channels for the delivery of SDCM information to users through geostationary and highly elliptical satellites. - Because of this work, the requirements for third generation SDCM satellite channels were approved and included in the technical specification currently being performed.

Research And Development Engineer

11.2008 - 09.2014 |Russian Space Systems

C++, MATLAB, python

- Hardware and software development for radio navigation, reliability calculation, satellite link budget calculation. Created a ground segment of SCDM satellite channels, including: - developed the hardware of the ground uplink stations (GUS) transmitting SDCM information to communication satellites; - developed the software to control the SDCM signal generators (C++); - developed the software for real-time monitoring and control the GUS equipment and analyze the results of GUS work (C++, Python); - managed the work on the manufacture, testing and deployment of three GUS.

Образование

Data Science professional training course

2020

Yandex School of Data Analysis

Radio Electronic Systems (Бакалавр)

2003 - 2009

The Bauman Moscow State Technical University

Языки

РусскийРоднойАнглийскийВыше среднего