About the Role
We are seeking a Senior Fraud Data Scientist to own the end-to-end lifecycle of fraud detection models — from research and prototyping, to production deployment, monitoring, and optimization. You will apply advanced machine learning techniques and engineering practices to fight complex fraud across payments, crypto, and trading.
This role requires a blend of data science and development skills: you’ll be coding, deploying, and optimizing fraud detection systems in real time. You’ll also collaborate with Fraud Ops, Product, Engineering, and Compliance to make sure our strategies are both effective and scalable.
Key Responsibilities
- Model Development: Research, design, and train ML models for fraud detection (supervised, unsupervised, anomaly detection, graph analysis, alert systems and red flag systems).
- Feature Engineering: Build fraud-specific features from transactional, behavioral, and blockchain data.
- Product ionization: Deploy ML models and scoring logic into production systems (batch + streaming) using Python and modern ML ops practices.
- Model Monitoring: Implement continuous monitoring pipelines for precision, recall, FPR, latency, drift, and false negatives.
- Optimization: Iterate on model improvements (feature tuning, hyperparameter optimization, retraining).
- Rule Integration: Combine machine learning with deterministic rules for hybrid fraud detection strategies.
- Lifecycle Management: Own retraining schedules, versioning, and model governance to ensure robustness over time.
- Collaboration: Work with engineers to optimize infrastructure, and with Fraud Ops to improve case routing and efficiency.
Requirements - 4+ years of experience in fraud detection, data science, or applied ML, ideally in fintech, payments, or crypto.
- Strong coding skills in Python (pandas, NumPy, scikit-learn, PyTorch/TensorFlow) and SQL.
- Proven experience building, deploying, and maintaining ML models in production.
- Familiarity with ML Ops workflows (CI/CD for models, monitoring, drift detection).
- Knowledge of anomaly detection techniques, classification models, graph-based ML, and network analysis.
- Exposure to real-time data pipelines (Kafka, Spark/Flink, or similar).
- Understanding of data engineering concepts (ETL pipelines, data warehousing, feature stores).
- Strong communication skills to align technical decisions with risk and business goals.
Nice to Have
- Experience with blockchain/crypto-specific fraud detection (wallet clustering, mixer detection, wash trading).
- Background in fraud operations or AML.
- Hands-on experience with containerization & deployment (Docker, Kubernetes).
Familiarity with streaming fraud detection frameworks or graph databases (Neo4j, TigerGraph).