Data Scientist
CurrentFuel supply forecasting for gas stations across the country:From Ad–Hoc tasks to product implementation, with segment expansion due to economic impact.1. Developed a forecasting framework with different levels of granularity (up to 10 thousand time series at a time) to reduce the company’s logistics business costs.2. Conducting experiments to refine models (CatBoost, Prophet, Chronos), tracking results (MLFlow), and analyzing deviations from business plans.3. Data extraction (SQL) and accounting for external factors (competitor prices, weather, downtimes) affecting the forecast.4. Packaging the project into a service with dashboards (Docker, Streamlit) for direct use by the client.5. Mentoring interns on tasks related to neural forecasting of time series (LSTM).
