Developing system for computing CPI using Scanner Data from scratch. ● Development and implementation of architecture on ClickHouse, Superset and Python. ● Development of an algorithm for the classification of product names (NLP). ● Development and implementation of CPI computing methodology based on scanner data.

Building a profitable FMCG business from scratch. Deep dive into all aspects of running SMEs. The CEO mindset was gained. Business development skills were obtained.

BTS Digital is a large IT company that develops many different B2C and B2B2C products on the local market, such as the messenger, video hosting, news aggregator, digitalId, educational projects, and many more. As head of the AI, my responsibility was to build and manage an AI team capable of delivering AI solutions in CV, NLP and Audio Processing for other products by demand. Our (actually multidisciplinary) team provided a full cycle of development of AI services from data gathering up to API. As a result: ● Developed face recognition algorithms are integrated into third party services and served already for more than 1 mln unique users in Digital ID project; ● Self STT for Kazakh and Russian languages for phone calls, TTS for Russian language was developed (including voice activity detection, language classification etc); ● Conversational chatbots for phone calls are developed; ● news duplicates detection and news classification algorithms are developed; ● Internal data annotation tool for audio and text data was developed. More than 100 hrs of audio data for ASR, more than 25 hrs audio of professional speaker in a professional studio for TTS were gathered; ● Many more ad hoc analytical and small ML tasks (adult content detection, market analysis based on scrapped data from Internet etc) are regularly conducted; ● A fast and continuously growing team and environment was provided. Internal and external ML seminars were organized.

As a head of the Data Science team of 5 people my responsibilities was to deploy BI and Data Science tools and make them work for the state-owned company with an enterprise culture and conservative environment. ● The results of the A/B test showed a 4% increase in sales ($2.8 million per year) due to the implemented bonus system (sales unit) based on monthly plans obtained from the developed forecasting algorithms and delivered to more than 3,500 thousand branches throughout Kazakhstan from the central office. ● The developed dashboards and the results of forecasting algorithms were used by the economic unit for annual planning, budgeting and tracking current results. The accuracy of planning has been increased and the number of monthly adjustments in the budget has been reduced. BI capabilities have allowed economists to analyze data at various scales interactively. ● Analytical ad hoc tasks for management, operational, economic and sales units were regularly solved

Conducting research in fields of topological data analysis and applied computational mathematics

Developing optimal algorithms for dumping DNS messages