Own models end-to-end across the fraud suite - transaction fraud detection, 3DS friction reduction, enrolment fraud, merchant fraud, cardholder risk scoring, and NLP/LLM-based products - from design through production deployment and monitoring.
Lead a team of 5–10 data scientists and engineers: set technical direction, review work, mentor junior members, and uphold engineering and modelling standards.
Drive feature discovery through deep EDA - form strong hypotheses, test them rigorously, and translate findings into production features that measurably reduce fraud and false declines.
Investigate client case studies : when banks report missed fraud or declines of genuine transactions, root-cause the issue in our pipeline, quantify impact, and propose model or feature fixes.
Build robust data pipelines and experiments using Python, complex SQL, and distributed/orchestration tooling.