Staff Data Scientist - Sam's Fraud Prevention

Duties: Architect and productionize Agentic AI-powered fraud decisioning system leveraging LLM orchestration, graph-derived embeddings, and ML-based risk scoring to aggregate multi-system signals, synthesize structured risk narratives, and generate automated accept/reject recommendations - materially reducing manual review dependency and decision latency. Design and deploy an ML-driven chargeback dispute automation agent that leverages structured feature extraction, document intelligence models, and evidence-ranking algorithms to auto-generate bank-ready dispute letters - improving recovery rates while reducing operational overhead. Build a unified cross-channel member profile by linking in-club, dotcom, returns, and transactional activity to create a holistic fraud-aware customer view, enabling more precise detection and reduced false positives. Develop and productionize an end-to-end batch inference model that complements the real-time ML pipeline by issuing delayed risk recommendations for high-risk transactions, strengthening overall fraud coverage. Engineer a large-scale distributed fraud graph platform leveraging GraphDB and GraphFrames to model multi-hop relationships across devices, payment instruments, addresses, and behavioral signals; implement Graph Neural Networks (GNNs) and unsupervised graph embeddings to detect coordinated fraud rings and synthetic identity clusters. Spearhead development of an intelligent retail AI system combining computer vision signals, behavioral ML, and anomaly detection algorithms to proactively prevent in-club theft - optimizing shrink reduction while preserving a frictionless customer experience. Establish feedback-driven fraud learning loops by integrating model outputs, manual review outcomes, and downstream signals into continuous system improvement. Own full ML lifecycle across multiple initiatives - from feature engineering and model design to deployment, monitoring, and business impact measurement. Partner cross-functionally with Engineering, Fraud Operations, Policy, and Retail stakeholders to operationalize advanced fraud systems at scale.

Minimum education and experience required: Master’s degree or the equivalent in Analytics, Economics, Computer Science, or related field OR Bachelor’s degree or the equivalent in Analytics, Economics, Computer Science, or related field plus 2 years of experience in analytics or a related field. Skills required: Must have experience with: coding in one of the following object-oriented programming languages: C++ or Python; Machine Learning and Deep Learning models, including CNN, Neural Networks and Bayesian Techniques; Statistics and Probability; performing data analysis and data collection using Python; identifying and applying metrics for measuring success and failure including F1, precision, recall and hypothesis testing; generating appropriate graphical representations of data and model outcomes; building scalable machine learning models for anomaly detection and Credit Risk; Data Visulaization in Python, Power BI or Tableau; AI Ethics, Model Fairness/Bias Monitoring; and Data and featuring engineering. Employer will accept any amount of graduate coursework, graduate research experience or experience with the required skills.