At Netformx, I built a client-facing sales analytics product from scratch, bringing together 7+ data sources, defining revenue KPIs, and shipping multi-dashboard reporting for customers.
I also built an entity-matching pipeline for Cisco hardware that improved identification accuracy by 30–40%, and designed a knowledge graph that powers a production AI assistant.
At Ingosstrakh, I owned a real-time incident risk model and validated it in a four-month A/B test on live traffic. It outperformed the existing process by 40% in monetary efficiency before rolling out to the full application flow.
At Sberbank, I used customer segmentation to personalize product offers, driving 11% more cross-sales of credit products. I also led a team of four master’s students through an ATM placement prediction project, from data collection and modeling to an interactive scoring service.
