• DSP (Demand Side Platform)– Spearheaded the development of bid optimization algorithms for a DSP in AdTech solution– Successfully reducing CPC(Cost per Click) by 25% and CPM(Cost per Mile) by 75% in remarketing campaigns through a series of experiments– Introduced results at webinar: https://www.youtube.com/watch?v=UR7ba3vMiN4• Virtual Gifts Ranking– Led the development of a ranking algorithm for virtual gifts in chats – Increased user engagement by 25% and virtual gift spending per paying user by 18%• DDA (Data-Driven Attribution)– Engineered and deployed a novel marketing campaign performance measurement algorithm using a Recurrent Neural Network– Achieved a 6% increase in eROI(expected Return on Investment) • Machine Learning for Direct Marketing– Developed a machine learning solution that optimized email campaign efficiency– Reduced email volume by 50% without compromising engagement (less than 5% drop in clicks), thereby preserving core business metrics• Landing Page Optimization– Initiated a continuous testing framework for landing pages using a Multi-Armed Bandit algorithm– Boosted conversion rates by 15% overall and up to 30% in some channels• Chat Copilot Implementation– Contributed to the team that implemented an LLM-driven chat copilot– Enhanced user interaction and operational efficiency. The project gained positive unit-economics since the first experiment iteration • LTV Prediction Models– Created and deployed multiple LTV(LifeTimeValue) prediction models– Increased marketing campaigns efficiency (up to +25% in eROI per campaign) via usage as predictive conversions

• Real-Time Scoring System– Engineered and maintained a real-time scoring system that includes a feature store and models service. My work encompassed integrating models, designing an additional abstraction layer for easier model integration by the data science team, implementing a model testing approach, and developing middleware for real-time ML monitoring– Significantly reduced time-to-market for new models and features, and improved the overall quality of the scoring system• Captcha Solver Algorithm– Led the development and enhancement of a Captcha solver algorithm that significantly increased data update frequency– Improved the relevancy of the credit scoring data by providing more frequent updates, which results in increased accuracy of scoring models• Revenue/Profit Prediction Algorithm– Created a predictive algorithm for businesses’ revenue and profits estimation– Integrated as a B2B service providing targeting data for financial products advertisement• Income Prediction Model– Developed and successfully integrated an individual’s income prediction model based on credit history data– Deployed as a profitable B2B solution• Custom Credit Scoring Models– Designed and developed bespoke credit scoring models for diverse B2B clients– Passed all on-site validations and were integrated into clients’ credit scoring pipelines• Machine Learning Utilities Package– Contributed significantly to the development of an internal machine learning utilities package– Greatly reduced time-to-market for new models and features development and research• Publication in Banking Journal– With a group of colleagues, published an article titled ”Risk-Factors - the Factors of Ineffective Credit Pipeline”

• Automation Tools Development– Developed and implemented a suite of automation tools for data collection in experimental environments– Significantly increased the number of samples gathered per a period of time, which allowed collecting datasets for statistical analysis • Experimental Data Analysis in Physics of Fluids– Designed and implemented various experimental setups on unique equipment stand. Collected and analyzed large sets of hydrophone data– Resulted in a publication and a conference presentation

• Uplift Modelling for Direct-Marketing– Led the development of an uplift model for promocodes, strategically targeting customers whose engagement would yield a positive margin– This approach increased Average Revenue Per User (ARPU) by 2.8% and improved margin per user by 3.2%• Customer Segmentation Insights– Provided the direct marketing team with crucial insights by employing customer clustering based on RFM (Recency, Frequency, Monetary value) and basket analysis features, enhancing targeted marketing strategies• Warehouse Digital Twin Development– Contributed to the creation of a digital twin model for the warehouse, enabling robust hypothesis testing and operational optimization• Recommender System Baseline– Developed a baseline solution for a recommender system, setting the stage for personalized customer experiences and increased engagement• A/B Testing for Data Science– Assisted in the design and analysis of A/B tests for the data science team, contributing to the validation and refinement of models and strategies through empirical data analysis