Senior Fraud Data Scientist — Fintech ML in Production

Circle; San Francisco (CA); On-site; USD 153,000 - 205,000; Full time

Job summary

Circle is seeking a Senior Data Scientist to own the prediction side of our fraud loop, turning labels into features and models, and measuring accuracy to move from vendor scores to in-house Circle models. You will collaborate with Fraud Strategy, Fraud Operations, Risk Engineering, and Compliance to design, backtest, and deploy production-grade fraud solutions.

You’ll define label taxonomy, set thresholds, monitor drift, and retrain models to stay ahead of evolving threats, protecting customers

Qualifications

• 5+ years in data science or applied ML on fraud, risk, or financial crime.
• Advanced Python and SQL, with XGBoost/LightGBM.
• Experience with sparse/biased labels and cold start.
• Strong statistics: anomaly detection and threshold optimization.
• Experience taking models to production and monitoring them.
• Clear communication with technical and non-technical partners.
• Ability to work through ambiguity and limited historical data.

Responsibilities

• Build and evaluate fraud models for onboarding, login, and money movement.
• Evaluate third-party risk signals and migrate to in-house models.
• Define label taxonomy with Fraud Operations and Risk Engineering.
• Design holdout and sampling to measure false negatives and blocking rate.
• Track fraud rate, dispute rate, and loss; set thresholds.
• Tune thresholds with Fraud Strategy and respond to attacks.
• Backtest, monitor drift, and retrain models on Circle’s platform.
• Support AI agent workflows for transaction review and fraud-ring analysis.

Skills

Python

SQL

XGBoost/LightGBM

Production ML

Communication

Ambiguity handling

Job description

Circle is seeking a Senior Data Scientist to own the prediction side of our fraud loop, turning labels into features and models, and measuring accuracy to move from vendor scores to in-house Circle models. You will collaborate with Fraud Strategy, Fraud Operations, Risk Engineering, and Compliance to design, backtest, and deploy production-grade fraud solutions.

You’ll define label taxonomy, set thresholds, monitor drift, and retrain models to stay ahead of evolving threats, protecting customers