Managing research projects.Topics: AutoML, Neural Architecture Search, Topological Data Analysis, Reinforcement learning.

Demand forecasting, dynamic pricing, predictive models for clicks/conversions/purchases, fuzzy products' descriptions matching with deep learning.

– Demand forecasting for X5 retail group (> 16’000 shops and >3500 products).∗Regular sales. Took the 1-st place in the data science competition with 5 industrial solutions for demand forecasting (IBM, Microsoft, SAP, JDA, Forecsys). The developed model was 12% more accurate than the nearest competitor.∗Promo sales. Developed the ML model for prediction of product sales during promotions. The model proved to be 19% more accurate than a customer's baseline.– Developed the ML-based recommender system for ”MediaMarkt” online shop. The system proved to be superior (+1.5% revenue, +3.2% conversion rate) over customers current baseline and better than the mature industrial solution (RichRelevance).

Research and development in Advertising Optimization Group. Research areas include: online advertising, CTR prediction, machine learning on computational advertising.Everyday work includes processing of Big Data with Map/Reduce cluster and planning of the live traffic experiments.

Statistical analysis of heat consumption in Moscow, automatic malfunction detection. Software engineering and research in power industry, systems integration, remote data acquisition.http://teplo-inform.ru

Software engineering and research in power industry, systems integration, remote data acquisition. Gas-drying plant modeling in Dynsim™ framework.