Sergej Odegov Email & Phone Number

Amsterdam, North Holland, Netherlands

Solving supplier fraud related problems with machine learning

Amsterdam Area, Netherlands

• Provided in-depth analysis and strategical outlook related to the payment fraud prevention product on new business verticals (Flights, Attractions, Ground Transportation).• Designed and built detection workflows and put relevant monitoring in place to spot anomalies and prevent any new fraud trends missed by ML models. Deployed rules to reject risky traffic or reroute to manual review.• Worked closely with engineering teams on new feature engineering and provided feedback on the general improvement of the fraud detection/prevention product.• Worked with fraud vendors to improve overall fraud performance, managed vendor relationships.• Supported senior management with creating relevant reporting on the team/department related KPIs.

Amsterdam Area, Netherlands

• Improved the capability of detection and prevention of fake partner listings by designing relevant reports to spot anomalies in the registration trends of new partners. • Have set the standards for the implementation of fraud rules to block risky registration traffic. Created multiple fraud rules to compliment the performance of the ML model and prevent any upcoming trends.• Mentored peers from the team and the department on improving their technical skills (SQL, pandas, pySpark).

Dublin, Ireland

Handled the financial recovery by generating response documents and responding to chargebacks, analyzing and limiting customer accounts in case of fraud or chargeback process abuse.

Vilnius, Lithuania

Analysed anomalies in payments traffic in order to identify new fraud trends/patterns. Provided in-depth analysis and insights to the Decision Science team with the purpose of improving the performance of ML models and rules.

Vilnius, Lithuania

Conducted phone interviews with customers sending money transfers to high risk destinations with a purpose of consumer fraud/scam prevention. Investigated customer transaction history in order to identify fraudulent activity.