Payment Fraud Machine Learning & Data Science Manager — Hybrid, New York, United States; San Francisco, United States

Overview

Lead hands-on fraud and payment-risk machine learning for high-volume bank-payment products. Build, ship and monitor models for ACH returns, fraud and account takeover while mentoring a team of data scientists and machine-learning engineers. Balance approval rates, customer friction, fraud losses and measurable financial outcomes in partnership with risk, product, engineering and finance.

Requirements

• At least five years of hands-on experience building and deploying machine-learning models in production, including two or more years leading a data science or machine-learning team.
• Current ability to build, validate, deploy and maintain production models personally, alongside leading others.
• Strong Python and advanced SQL working with large datasets in cloud data warehouses or distributed processing engines.
• Direct experience modeling fraud, credit or payments risk, including rare events, delayed outcome labels, reject inference and business impact beyond AUC.
• Experience with real-time model inference and monitoring for high-volume, low-latency risk decisions, including drift detection, retraining and updates.
• Experience owning or materially contributing to loss forecasting or a loss budget and explaining financial impact to stakeholders.
• Strong experimentation for model and policy changes, including champion-challenger evaluations, holdouts and staged rollouts.
• Ability to mentor data scientists and machine-learning engineers and explain risk-model tradeoffs to nontechnical leaders.
• A quantitative degree; scikit-learn, XGBoost, LightGBM or CatBoost; ACH/NACHA knowledge; open-banking data; feature stores; graph/sequence models; cloud ML platforms; and model-risk governance experience are preferred rather than mandatory.

Responsibilities

• Personally build, validate, deploy and improve production fraud and ACH-return risk models.
• Lead and mentor a data science and machine-learning engineering team and set its delivery roadmap.
• Engineer transaction, device, account, network and behavioral features for low-latency decisioning.
• Measure model and policy changes using controlled experiments, backtests and champion-challenger comparisons.
• Own monitoring of model performance, drift, loss drivers and operational outcomes.
• Partner with risk, product, finance and engineering to connect scoring decisions to customer approvals and financial losses.
• Document model decisions, validation, governance evidence and explainable outcomes for auditors and internal reviewers.
• Evaluate external risk signals and data providers for incremental fraud-detection value.

Conditions

• Full-time hybrid role based in New York or San Francisco, United States, with office attendance Tuesday through Thursday.