Skip to content

Alexandr Belykh

@alexandr-belykh

Data Scientist & ML Engineer

Data Scientist with one year of commercial experience in telecom and FMCG. IITU graduate (Data Science, GPA 3.73 / 4).

Open to

Learning with othersFinding a mentorJoining competitions & hackathonsCompany partnerships

About

Currently a Data Scientist at Philip Morris International. I rebuilt and automated the market share forecasting model for newly launched cigarettes, reducing MAE from 63% to 3.4%, and developed a classification model for selecting retail outlets for sales boost campaigns (PR-AUC 64%), which saved 15% of the campaign budget. Previously at Kcell (Big Data), I built an SQL solution for finding subscribers with a second SIM from another operator based on behavioral data (for more targeted advertising) and designed the logic of a model that scores good vs. bad taxpayers among small and medium businesses to offer them targeted tariffs. Analytics background: at Netcracker I designed data marts (catalog, CPQ), wrote FDS/DDS documentation, built Apache Superset dashboards and took part in deployment automation with Kubernetes, Git and Jenkins. At Kazakhtelecom I analyzed billing and survey data and built Qlik Sense dashboards, helping cut the share of debtors from 30% to 16% and attract new customers (+8%). Personal projects: credit scoring (with VAE data augmentation and Streamlit deployment), automated IELTS essay scoring, and Krisha.kz scraping with weekly automated analytics. Tech: Python, SQL, PyTorch, scikit-learn, XGBoost, CatBoost, LightGBM, Pandas, PySpark, Hive, Impala, Snowflake, Oracle, PostgreSQL, Power BI, Power Query, Qlik Sense, Docker, Git, Jira, Kanban. Interested in LLMs, RAG and GenAI integration; regularly take part in hackathons and follow AI/ML research. Quick learner, used to working in product teams alongside analysts and developers. Aiming to grow as an ML/DL/NLP Engineer.

Career

Experience: 1 years

Experience

Data Scientist

Philip Morris International

A global tobacco company with operations across many markets, known for cigarette brands and a growing portfolio of smoke-free products.

Aug 2026 — Present

Rebuilt and automated a new cigarette market share forecast, cutting MAE from 63% to 3.4%, and built a campaign targeting model that saved 15% of budget.

• Improved the market share forecasting model for newly launched cigarettes (NPL CC Model): redesigned and rebuilt the model logic, reducing MAE from 63% to 3.4%, and automated the model pipeline. • Developed a classification model to select retail outlets for sales boost campaigns (PR-AUC 64%), which saved 15% of the campaign budget.

Data Scientist

Kcell

A major Kazakhstan mobile operator providing consumer and business telecom services under the Kcell and Activ brands, along with digital services.

Jun 2026 — Jul 2026

Created SQL logic to identify subscribers using a competitor's second SIM and designed a model to segment small and medium business taxpayers for targeted tariffs.

• Wrote an ad-hoc SQL solution that identifies subscribers who use a second SIM of another telecom operator based on their behavioral data, enabling more targeted advertising. • Designed the logic of a model that identifies potentially good and bad taxpayers among small and medium businesses, in order to offer them targeted tariffs.

Data Analyst

Netcracker Technology

A global software and services provider whose platforms support telecom operators' business operations, networks, and digital transformation.

Feb 2026 — May 2026

Created and updated data marts, wrote documentation, and built KPI dashboards, with reported gains in system efficiency, data flow, and revenue and sales accuracy.

• Created data marts (catalog, CPQ) and wrote FDS and DDS documentation, which improved system efficiency by 13.7%; updated existing data marts, optimizing data flow by 6%. • Built Apache Superset dashboards with documentation and took part in the automation and deployment cycle (Kubernetes, Git, Jenkins) for KPI visualizations, resulting in a 16% improvement in revenue and sales accuracy by internal metrics.

Data Analyst

Kazakhtelecom

Kazakhstan's national telecommunications operator, providing fixed-line, mobile, broadband, and related digital services across the country.

Jun 2024 — Aug 2024

Analyzed billing and survey data, built debt tracking dashboards, and supported campaigns, with reported reductions in debtors and an increase in new customers.

• Analyzed billing data using SparkSQL and built Qlik Sense dashboards to track customer debts, helping to reduce total debt volume by 2% and cut the share of debtors from 30% to 16%. • Worked with user survey data and took part in promotional campaign development, which increased the number of new customers by 8%; updated data marts in PostgreSQL.

Education

Data Science

IITU (International Information Technology University)

Selected projects

Credit scoring model

Built a model to predict loan approval probability and deployed a user interface for it.

Automated IELTS essay scoring system

Developed an automated system for scoring IELTS essays.

Skills

PythonSQLC++PandasNumPyPyTorchTensorFlowScikit-learnXGBoostCatBoostLightGBMHuggingFacetransformersPySparkMatplotlibSeabornSeleniumPostgreSQLOracleSnowflake