At Sky Deutschland, I owned four machine-learning data products used by more than 6M customers: a next-best-action decision engine, upsell propensity models, intelligent customer routing, and AdSmart addressable TV. Bringing model recommendations into agent, web, CRM, and campaign channels raised call conversion by 20% and digital upsell by 11–16%.
I reached 82% accuracy in intelligent customer routing and used A/B tests to compare routing rules. I also introduced a rotation algorithm to reduce offer fatigue and set up monitored scoring pipelines with GDPR-compliant data integration into DataHub.
Before Sky Deutschland, I worked in commercial analytics at A1 Hrvatska, data governance and stress testing at Addiko Bank, and forecasting at Henkel Slovensko. At Henkel, I developed a Python tool for intermittent sales data that saved more than three FTE days per week.

