Semyon Shafronov

• Achieved hundreds of successful deals and generated LTV revenue of hundreds of thousands of euros by developing a model to predict customer propensity to purchase B2B products. • Improved the lead-to-deal conversion rate from 5% to 7% by addressing model overfitting, correcting target imbalance, and expanding feature sources (utilizing petabytes of data from a corporate data lake, open sources, and three affiliate companies). • Reduced model update labor costs from 2 days to 2 hours through automation.

• Achieved an n-percent decrease in shadow transactions by implementing 2 ML-models and 20 complex rules for identifying suspicious transactions within a high-load system that processes 500,000 online payments daily. • Reduced labor costs for AML department employees involved in post-analysis of client transactions and report preparation for regulators by developing 65 complex reports featuring filters, dictionaries, ETL tools, execution queues, conditional mailings, and role-based access (85 daily users).