• Improved reliability of the personalization system by developing a robust recovery of a real-time state from the petabytes of the search log;• Increased search quality by 1% by developing a real-time metric evaluation system.• Expanded the field for experiments by developing a URL profile on behavioral factors.• Four junior developers under my leadership are implementing various big data processing pipelines.

I continued to advise the company on working with data, maintaining the code, and mentoring new employees. Integrated the company's projects into work with students at the university to the benefit of all participants.

• Designed and developed a matching system for 25,000+ products which gave the company a decisive advantage to overcome the break-even point;• Started with the classification of short texts by training narrow-domain Word2Vec combined with KNN1. In a few iterations, using hierarchical clustering and a trie-like data model, I got a rule-based solution that surpassed initial expectations;• Established a procedure for marking up data and offline evaluation of the model;• Developed a sales forecasting for each product of 10000+ points of sale;• Compiled recommendations for offline sales with an increase in revenue by at least 8%.