Skills Required: This position requires two (2) years of experience with the following: developing and deploying end-to-end supervised and unsupervised ML models, including Decision Trees, XGBoost, LightGBM, and K- means, for fraud detection in the financial services or payments industry; working with high throughput real-time transactional data at scale; using Docker, Kubernetes, and CI/CD pipelines to deploy models such as gradient boosted trees or deep learning architectures; developing graph- based ML solutions for fraud detection with GNNs using GraphSAGE, node2vec, metapath2vec, or Graph Attention Networks; leveraging Pytorch Geometric or NetworkX; creating risk scores using temporal features, rolling aggregates, and longitudinal modeling to support fraud prevention KPIs; building distributed data pipelines for feature engineering and model training using PySpark, Apache Beam, Kafka, and Airflow; building data warehouses that focus on feature freshness and low latency using stacks that leverage BigQuery or Snowflake; using Python, TensorFlow, PyTorch, Scikit-learn, Keras, XGBoost, and LightGBM for fraud detection model development; implementing model fairness, explainability in SHAP and LIME, and compliance in a regulated financial environment; using model governance in the financial industry, including model risk management reviews, compliance documentation, and responding to audits; working with high-cardinality categorical features in embeddings and statistical smoothing for merchant-level behavior modeling or user device fingerprinting; conducting exploratory data analysis on large-scale, high-dimensional datasets; identifying signals in noisy transaction data, uncovering fraud patterns, and informing feature engineering and modeling decisions; extracting, transforming, and analyzing data from structured financial databases using advanced SQL techniques, including complex joins, subqueries, common table expressions, window functions, and stored procedures; performing scalable data processing using PySpark, BigQuery, Dask, and visualization of distributions; performing time series analysis using matplotlib, seaborn, and Plotly under compute and memory constraints. Engineer graph-based features and embeddings by constructing transaction-level payment graphs across cross-functional teams and applying Graph Neural Networks (GNN) to generate features for fraud detection models.