Built and deployed credit risk and fraud detection models using Python, PySpark, XGBoost, and LightGBM on Azure Databricks, processing over 30 million monthly transactions across consumer banking portfolios. Developed scalable feature engineering pipelines leveraging Spark, Kafka, Delta Lake, and Snowflake to generate risk signals for production ML models, reducing feature generation latency to under one hour. Designed graph-based fraud detection solutions using NetworkX and entity resolution to identify synthetic identities, mule accounts, and coordinated transaction networks. Operationalized end-to-end model lifecycle management through MLflow, Model Registry, automated CI/CD pipelines, drift monitoring, and champion-challenger validation frameworks to improve production governance and deployment reliability. Collaborated with Credit Risk, Compliance, Fraud Strategy, and Collections teams to build customer segmentation, propensity, and early-risk models, improving targeting of high-risk customer segments. Built retrieval-augmented knowledge assistants using Azure OpenAI, LangChain, embeddings, and vector search to support compliance policy navigation and analyst research workflows, reducing manual knowledge retrieval efforts. Implemented explainability and Responsible AI controls using SHAP, LIME, fairness validation, model documentation, and audit workflows to strengthen model risk management and support regulatory review processes. Developed Power BI reporting frameworks for monitoring portfolio risk trends, model drift indicators, prediction stability, and operational KPIs, enabling proactive model maintenance and faster issue identification across stakeholders.…Built and deployed credit risk and fraud detection models using Python, PySpark, XGBoost, and LightGBM on Azure Databricks, processing over 30 million monthly transactions across consumer banking portfolios. Developed scalable feature engineering pipelines leveraging Spark, Kafka, Delta Lake, and Snowflake to generate risk signals for production ML models, reducing feature generation latency to under one hour. Designed graph-based fraud detection solutions using NetworkX and entity resolution to identify synthetic identities, mule accounts, and coordinated transaction networks. Operationalized end-to-end model lifecycle management through MLflow, Model Registry, automated CI/CD pipelines, drift monitoring, and champion-challenger validation frameworks to improve production governance and deployment reliability. Collaborated with Credit Risk, Compliance, Fraud Strategy, and Collections teams to build customer segmentation, propensity, and early-risk models, improving targeting of high-risk customer segments. Built retrieval-augmented knowledge assistants using Azure OpenAI, LangChain, embeddings, and vector search to support compliance policy navigation and analyst research workflows, reducing manual knowledge retrieval efforts. Implemented explainability and Responsible AI controls using SHAP, LIME, fairness validation, model documentation, and audit workflows to strengthen model risk management and support regulatory review processes. Developed Power BI reporting frameworks for monitoring portfolio risk trends, model drift indicators, prediction stability, and operational KPIs, enabling proactive model maintenance and faster issue identification across stakeholders.WWWW…