Developed an algorithm to extract socially significant products from fiscal raw data, which was used by the Kazakhstan Government as an alternative price monitoring system; Developed a telegram-bot with a predictive model, which increased the labeled data ] 10 times and the accuracy of the main model by 5%; Created a classification model that determines the category, brand, and other valuable information from fiscal raw data; Processed 100 mln rows worth of data (or 35 mln cashier receipts) on a weekly basis.

Developed a machine learning model for new building's starting prices, which increased the company's annual revenue by 500 million tenge (approx. $1.7 mln); Developed a dynamic pricing algorithm for different flat types (multi-armed bandits), which lead to steady sales of liquid and non-liquid flats; Used geospatial data for building price predictions (house prices, criminal stats, business stats) from external sources, which increased the model accuracy by 10%.

Developed statistical and econometric models for country-level decisions.

Developed statistical and econometric models for the manufacturing industry (on Eviews); Identified Kazakhstan’s priority goods in the international market using the gravity model, which was utilised on the government level; Determined factors that influence the export of products by the regression models.

Developed statistical and econometric models for Kazakhstan’s economy.