Engineered hybrid CNN–Transformer pipelines with TensorFlow and Keras across 4M laparoscopic frames, enabling surgical phase classification with contextual embeddings and HIPAA compliance. Built ETL workflows to preprocess 2.3TB of imaging and clinical text streams into structured formats for multimodal DL. Designed anomaly detection frameworks using GANs and reinforcement learning to proactively identify rare intraoperative complications while maintaining confidentiality. Deployed optimized inference services with Flask and TensorRT, exposing low-latency APIs into surgical dashboards for predictive overlays and decision-support across EHRs. Enhanced temporal forecasting with RNNs and Bayesian Optimization, reducing latency for surgical activity prediction across 30 concurrent feeds. Scaled pathology AI workflows on AWS SageMaker and EC2 with DICOM-compliant pipelines and MLflow monitoring for reproducibility.

AI/ML Engineer at Deloitte

July 1, 2023 - July 1, 2023

Developed credit scoring pipelines using XGBoost and Scikit-learn on 15M retail loan applications, improving default prediction accuracy by 11% while supporting Basel III compliance. Orchestrated batch processing with Apache Spark and Hadoop on 3TB transactional data for fraud detection analytics. Engineered supervised forecasting models with LightGBM for time-series repayment histories, reducing delinquency forecast errors by 17%. Built Power BI dashboards for regulatory teams and automated cleansing of 4.5TB cross-border payments data using Informatica ETL. Deployed credit risk APIs with Django and SQL Server enabling real-time scoring for 120,000 applications per month within PCI DSS. Implemented model monitoring with Azure Databricks and drift analysis, reducing false positives by 9%. Containerized apps with Kubernetes and CI/CD via Jenkins, cutting deployment effort by 31%.