Staff ML Risk Analytics

What you'll be doing

• Define the ML data and feature strategy for fraud detection, determining what data needs to enter our systems so our models can take intelligent, high-accuracy action on a small fraction of traffic where intervention matters most.
• Own the end-to-end feature engineering pipeline identifying, building, validating and promoting features that drive measurable improvements in ATO and scam ML performance.
• Diagnose gaps between current tooling infrastructure and the solutions needed, and drive the roadmap to close them leveraging your understanding of how the industry has evolved to make the right architectural calls.
• Partner with Machine Learning Engineers to translate analytical insights into production-ready ML systems, ensuring models are instrumented, monitored, and continuously improved.
• Set technical direction for the ML Analytics function within Growth & Risk, mentoring junior team members who need a senior practitioner to define the approach and translate direction into execution.
• Partner cross-functionally with Product Managers and Risk analysts to surface fraud signals and translate ML findings into business-impacting decisions.
• Serve as the team's institutional knowledge resource on ML industry evolution — helping the organization understand why certain solutions work, what historical architectural decisions mean for current tooling, and where the industry is headed next.

What we look for in you

• 8+ years of hands-on experience in machine learning analytics, data science, or a related technical field with meaningful experience applied to risk, fraud, or payments problems.
• Deep, practitioner-level expertise in Spark, Python, and big data ML this is the core stack. SQL and rule-writing are adjacent skills; they are not what this role is about.
• Proven experience in feature engineering for ML models, including identifying the right signals, building pipelines, and validating feature quality at scale.
• Holistic understanding of how the ML industry has evolved over the past decade from Hadoop-era big data to modern feature stores like Tecton and the ability to apply that knowledge to close infrastructure gaps.
• A curated, high-precision approach to ML problems: you understand that in fraud and risk, you are optimizing for sensitivity and accuracy on a small fraction of traffic.