Fraud Data Scientist

About the role

As a Fraud Data Scientist, you will be at the front lines of protecting our ecosystem from sophisticated financial fraud and abuse. You will join a high‑impact team operating in a data‑rich, high‑frequency environment where seconds matter.

Key responsibilities

• Design, train, and deploy advanced machine‑learning models (gradient boosting, anomaly detection, graph networks) for real‑time fraud detection.
• Own the full production lifecycle, ensuring low‑latency execution and high reliability of models in a high‑traffic environment.
• Research and implement agentic flows and LLM‑driven orchestration to automate multi‑step fraud decisioning and investigation.
• Conduct deep‑dive exploratory analysis on massive datasets to uncover emerging fraud vectors and coordinated attacks.
• Build and optimise real‑time streaming and batch features to improve model signal and precision.
• Design shadow‑testing and A/B testing frameworks, and set up continuous monitoring pipelines to detect data drift and performance degradation.

Required profile

• Minimum 3 + years of applied data‑science experience, preferably in fintech, fraud, risk or payments.
• Proven hands‑on experience deploying and maintaining machine‑learning models in high‑traffic, real‑time production environments.
• Strong ability to translate model performance into business impact, balancing precision/recall with financial and operational considerations.

Required skills

• Python (Pandas, NumPy, Scikit‑Learn, XGBoost/LightGBM)
• Advanced SQL for querying massive, complex datasets
• Experience with cloud data platforms such as Databricks and data warehouses like BigQuery
• Familiarity with MLOps/orchestration tools (MLflow, Airflow)