Lead Data Scientist/Analyst

• Responsibilities - Data Analytics and Data Visualization of Yandex Cloud data. - I developed concept new ML based antifraud solution and event monitoring system (with BigData). - Developed ML scoring model to find potentially profitable consumers. - Made resources demand forecasting tool. - Created ID graph of clients. • Impact - New antifraud system allows to detect more fraud than the set of rules and automatically adapts for new conditions. - BI dashboards allows to easily estimate current status and get new insights from data. - Upsale model analyzes consumer behavior and extracts patterns of profitable consumers and also reduces human labor. - Resource demand forecasting tool allows to be sure that users will have enough resources. - Based on ID graph clients are matched with organizations from external datasources (community detection). This information is used in scoring model.

• Responsibilities - Leaded data analytics team. - Generalized burst detection algorithms from water to oil&gas use cases. - Developed transient modeling algorithm and corresponding surrogate neural networks model. - Water Quality Analysis (collaboration with colleagues from UK) • Impact - Results of development allows to tune burst localization algorithm without real life experiments. - For water quality analysis was the report was created to provided initial estimations for further project development.

In this period I participated in the burst detection project as the lead data analyst. This project was developed in collaboration with colleagues from UAE. • Responsibilities - Developed algorithms for water burst detection, localization and classification of events (PoC). - Generalized algorithm for water supply networks of arbitrary size and complexity. - Made it scalable relatively number of sensors (GPGPU was used). - Developed analytics module from PoC to Version 1 for production environment. • Impact - Production ready analytics module for burst detection was developed.

• Responsibilities Gas turbine projects in collaboration with colleagues from Germany: - Created new neural networks based algorithm to improve turbine's blades optimizer. - Developed neural network to predict blade's Von Mises stress. - Created PoC application to predict scrap rates for blades and vanes in gas turbines. Acted as project manager in scrap rates prediction project (planning and understand the customer’s business needs) • Impact - Algorithm for blades and vanes design optimization improved existing optimizer and improved computational performance. - Von Mises stress prediction model allows to replace physical modeling by ML, what is 10 times faster. - Found existing data problems for Scrap Rates PoC. This PoC become the point of collaboration between 3 departments, which didn't know about common problems. ML model not only predicts, but also allows to find outliers and physical dependencies.

Full time internship during the second year of the magistracy • Responsibilities - Created BI reports for water supply monitoring system and improved their evaluation speed in more than 10 times. - Created Java endpoints to improve performance and added new functionality. - Wrote a few invention disclosures for water supply systems. - Created masters thesis in collaboration with Siemens about applications of Game Theory (Minimization of water consumption peak) and Machine Learning (Forecasting of water consumption for non automated flow meters) to water supply systems. • Impact - As intern I saved for Siemens a lot of money (because they didn't need to hire FTEs) without quality losses. - Customer became interested in my master thesis. - Because of performance improvements user experience became significantly better.

During the first year of the magistracy. • Responsibilities - Analysis of medical statistics. - Automated development of regular reports with Microsoft SQL Server Reporting Services. • Impact - Automated reports reduced human labor in more than 7 times.

Development of reports with Microsoft SQL Server Reporting Services.