Roman Prilepskiy
Lead ML developer - MTS AI at ARTIFICIAL INTELLIGENCE CENTER- Claim this Profile
Click to upgrade to our gold package
for the full feature experience.
-
English Professional working proficiency
-
Russian Native or bilingual proficiency
-
Italian Elementary proficiency
Topline Score
Bio
Ivan Isaev
I worked with Roman at the pull of data engineering and machine learning projects for operational efficiency improvement and process mining. Roman combines advanced developer and ML engineer skills with a team leader trait who can organize effective team work. Roman hard skills are very good in scopes of backend development, NLP and predictive analytics. He also can successfully work and adapt in quickly changeable environment and highly involved and interested in successful team work and achieving of global project goals.
Evgeny Rebrin
Работали с Романом над проектом по созданию игрушки для детей в FabLab НИТУ «МИСиС». Проект оказался нелегким, задумка была весьма нестандартная и благодаря находчивости и смекалки Романа, нам удалось найти необходимые инструменты и оборудование для ее реализации. За время работы я отметил следующие качества Романа: находчивость, ответственность, умение работать в команде, инициативность и нацеленность на результат. По моим наблюдениям, у Романа есть задатки эффективного и результативного руководителя.
Ivan Isaev
I worked with Roman at the pull of data engineering and machine learning projects for operational efficiency improvement and process mining. Roman combines advanced developer and ML engineer skills with a team leader trait who can organize effective team work. Roman hard skills are very good in scopes of backend development, NLP and predictive analytics. He also can successfully work and adapt in quickly changeable environment and highly involved and interested in successful team work and achieving of global project goals.
Evgeny Rebrin
Работали с Романом над проектом по созданию игрушки для детей в FabLab НИТУ «МИСиС». Проект оказался нелегким, задумка была весьма нестандартная и благодаря находчивости и смекалки Романа, нам удалось найти необходимые инструменты и оборудование для ее реализации. За время работы я отметил следующие качества Романа: находчивость, ответственность, умение работать в команде, инициативность и нацеленность на результат. По моим наблюдениям, у Романа есть задатки эффективного и результативного руководителя.
Ivan Isaev
I worked with Roman at the pull of data engineering and machine learning projects for operational efficiency improvement and process mining. Roman combines advanced developer and ML engineer skills with a team leader trait who can organize effective team work. Roman hard skills are very good in scopes of backend development, NLP and predictive analytics. He also can successfully work and adapt in quickly changeable environment and highly involved and interested in successful team work and achieving of global project goals.
Evgeny Rebrin
Работали с Романом над проектом по созданию игрушки для детей в FabLab НИТУ «МИСиС». Проект оказался нелегким, задумка была весьма нестандартная и благодаря находчивости и смекалки Романа, нам удалось найти необходимые инструменты и оборудование для ее реализации. За время работы я отметил следующие качества Романа: находчивость, ответственность, умение работать в команде, инициативность и нацеленность на результат. По моим наблюдениям, у Романа есть задатки эффективного и результативного руководителя.
Ivan Isaev
I worked with Roman at the pull of data engineering and machine learning projects for operational efficiency improvement and process mining. Roman combines advanced developer and ML engineer skills with a team leader trait who can organize effective team work. Roman hard skills are very good in scopes of backend development, NLP and predictive analytics. He also can successfully work and adapt in quickly changeable environment and highly involved and interested in successful team work and achieving of global project goals.
Evgeny Rebrin
Работали с Романом над проектом по созданию игрушки для детей в FabLab НИТУ «МИСиС». Проект оказался нелегким, задумка была весьма нестандартная и благодаря находчивости и смекалки Романа, нам удалось найти необходимые инструменты и оборудование для ее реализации. За время работы я отметил следующие качества Романа: находчивость, ответственность, умение работать в команде, инициативность и нацеленность на результат. По моим наблюдениям, у Романа есть задатки эффективного и результативного руководителя.
Credentials
-
How to Make Product and Business Decisions Based on Metrics
Heroes.campAug, 2023- Oct, 2024 -
Building Conversational Experiences with Dialogflow
Coursera Course CertificatesJul, 2019- Oct, 2024 -
Machine Learning Foundations: A Case Study Approach
Coursera Course CertificatesJun, 2016- Oct, 2024 -
Математика и Python для анализа данных
Coursera Course CertificatesMay, 2016- Oct, 2024 -
Комплексное обучение. Работа в ЕМИАС
Сетевая академия ЛАНИТApr, 2016- Oct, 2024 -
Введение в машинное обучение (Introduction to machine learning) - Higher School of Economics
CourseraMar, 2016- Oct, 2024 -
Exploratory Data Analysis
Coursera Course CertificatesFeb, 2016- Oct, 2024 -
Kaggle R Tutorial on Machine Learning
DataCampJul, 2015- Oct, 2024 -
Machine Learning (Grade Achieved: 96.8%)
CourseraApr, 2015- Oct, 2024 -
Getting and Cleaning Data
Coursera Verified CertificatesFeb, 2015- Oct, 2024 -
R Programming
Coursera Verified CertificatesFeb, 2015- Oct, 2024 -
The Data Scientist’s Toolbox
Coursera Verified CertificatesFeb, 2015- Oct, 2024 -
An Introduction to Interactive Programming in Python
CourseraNov, 2014- Oct, 2024
Experience
-
ARTIFICIAL INTELLIGENCE CENTER
-
Indonesia
-
Education Management
-
1 - 100 Employee
-
Lead ML developer - MTS AI
-
Jan 2020 - Present
Role: DS Team Lead (5 people) Automated operational efficiency system - Managed Scrum-based development of an automated operational efficiency system which monitored 4 large departments supporting MTS Mobile Infrastructure. Project was successfully approved on MTS Product Committee. - Scope included database development, ETL with updates every night, KPI development, REST API, PowerBI dashboard, forming a departments' competency matrixes, workload matrixes, peak staff workload; - Technology stack used: Flask, Flask blueprints, SQLAlchemy ORM, PostgreSQL, psycopg2, cx_Oracle, cron, Docker, Docker-compose, PowerBI, pytest, alembic, uwsgi, asyncio, mypy, black, pylint, flake8. Base station downtime calculation - Developed a FastAPI web-service to calculate downtime (in min) of telecom base stations during planned maintenance works; - Managed development of large data storage for accumulating historical data. Before one of the data sources stored actual data only for 40 days. Role: ML Developer - Developed a Flask-based spellchecking microservice with ML-model's quality (WER and ACC) on domain telecom data better than any solutions used before; - Created a custom pipeline component - spellchecking - based on RASA chatbot platform, with REST API usage connecting to a separate microservice and a Telegram-bot for testing; - Conducted a competitive research on modern and SOTA approaches to spellchecking. Results included: GitHub repos (8 in total), literature review (arXiv, Semantic Scholar - 118 in total), patents (Yandex, Google, Microsoft, IBM, Baidu, Samsung - 151 in total). Selected most promising ideas to implement; - Initiated development of a batch-based testing API for spellchecker based on current production platform; - Negotiated with an external development team to give us a long trial of their spellchecking enterprise solution for free; - Developed several ML prototypes of spellchecking models (Transformer, RuBERT, YouTokenToMe, SymSpell). Show less
-
-
-
TerraLink
-
Chile
-
Renewables & Environment
-
1 - 100 Employee
-
Middle Data Scientist
-
Jan 2019 - Dec 2019
- developed a text classification model for classifying service request to IT Support department in a telecom company (F-score = 0.818, 65+ categories); - participated in presales of the ML solutions (including planning the project workload to calculate the budget, presenting ML part of the solution); - conducted a competitive analysis of ML consulting firms in Russia, 70+ companies included (business profile, SEO, financial data, court cases, number of employees); - performed ML skills & experience audit interconnecting results with TerraLink's previous and current clients to propose new ML solutions to build and sell; - created 5 presentations (Rus, Eng) on TerraLink's Data Science department, its solutions and cases; - created a detailed Value Proposition Canvas on one of the ML Solutions to make a landing page and start a targeted marketing campaign; - proposed and participated in interviews with other TerraLink's practices, departments on their clients, "pains" and problems, possible collaboration on projects. Show less
-
-
-
BestFit|Car
-
Russian Federation
-
Information Technology & Services
-
Data Scientist
-
Aug 2017 - Dec 2018
- Developed a new recommendation engine for the company's car recommendation system, ~5x faster; - Conducted 10 A/B tests (web-based) to tune the car recommendation system covering ~500 people, which led to 3x increase in percentage of satisfied customers; - Proposed and developed a "like" prediction model - whether a customer will like a car recommendation or not (f-score = ~0.75) - based on the feedback data from A/B tests; - Implemented a person-to-person matching model (by psychological "fitness"); - Parsed and preprocessed ~15 000 articles about cars from 10 web-sources; - Created a classification model to predict an article type (accuracy = ~0.8, f-beta score = ~0.7); - Prepared tech part (data, task, metrics) and orally presented a Kaggle inclass challenge from our company for the hackathon "AI.Hack Moscow", judged our case (26 teams) and others. Show less
-
-
-
Data Analysis Center
-
United States
-
Oil and Gas
-
1 - 100 Employee
-
Data Analyst
-
Jun 2015 - Aug 2017
Data Analyst: - Developed PL/SQL scripts to analyze EMIAS (United Medical Information and Analytical System of Moscow) on different cases. - Designed ETL process and PL/SQL scripts to build an internal DWH. - Implemented a IPython-module with ETL process, PL/SQL scripts, data cleaning and feature engineering to build a detailed clinic's profile. - Inspected Moscow municipal hospitals for anomalies in usage dynamics of different appointment channels to doctors. - Examined availability of doctors for patients and patients' feedback in 2015-2016 for the Annual Moscow Healthcare report for S. Sobyanin. - Conducted competitive analysis of the national healthcare systems in Japan and Germany. - Designed a set of fuzzy logic algorithms for finding and eliminating patients' doubles in several EMIAS databases. - Built a machine learning model for forecasting final half-year and year marks of children on school subjects. Team Lead: - Coordinated competitive and functional analysis of worldwide web-services and applications in the field of electronic personal health management. - Developed validation process (system analysis, ETL process, PL/SQL scripts) of results of patients' data migration between two databases. Show less
-
-
-
Intel Corporation
-
United States
-
Semiconductor Manufacturing
-
700 & Above Employee
-
Intern
-
Jun 2014 - Jul 2014
- Analyzed the current and future IoT market situation. - Interviewed IoT-relevant Russian companies on existing problems and constraints. - Introduced a possible Intel’s strategy of overcoming the IoT market fragmentation and boosting Intel's IoT business. - Analyzed the current and future IoT market situation. - Interviewed IoT-relevant Russian companies on existing problems and constraints. - Introduced a possible Intel’s strategy of overcoming the IoT market fragmentation and boosting Intel's IoT business.
-
-
-
-
Programmer
-
Apr 2012 - Jul 2013
- Developed PL/SQL scripts for the air traffic management system of the Russian Federation. - Designed UI in QT for an in-house Windows-client of the database. - Developed PL/SQL scripts for the air traffic management system of the Russian Federation. - Designed UI in QT for an in-house Windows-client of the database.
-
-
-
Shuffle
-
United States
-
Advertising Services
-
Consultant
-
Aug 2011 - Apr 2012
Responsibilities: 1. Business processes and documentation analysis; 2. Software functional requirements acquisition and analysis; 3. Designing models of the web-services in Microsoft Visual Web Developer. Achievements: 1. Took part in executing state project - The System of Interagency Electronic Interaction - for the Federal Air Transport Agency through development of technical documentation for Information System. Responsibilities: 1. Business processes and documentation analysis; 2. Software functional requirements acquisition and analysis; 3. Designing models of the web-services in Microsoft Visual Web Developer. Achievements: 1. Took part in executing state project - The System of Interagency Electronic Interaction - for the Federal Air Transport Agency through development of technical documentation for Information System.
-
-
Education
-
2013 - 2015
Skolkovo Institute of Science and Technology
Master of Science (MSc), Information Science and Technology -
2013 - 2015
Moscow Institute of Physics and Technology (State University) (MIPT)
Master's Degree, Intellectual Data Analysis -
2013 - 2013
Massachusetts Institute of Technology
Workshop, Innovation and Entrepreneurship -
2007 - 2013
Bauman Moscow State Technical University
Engineer's degree, Design and Manufacturing Technology of Electronic Devices
