AI/ML Engineer at Northern Trust

Designed and deployed an enterprise-grade Generative AI platform using LangChain, LlamaIndex and Azure OpenAI to enable intelligent search and summarization across investment reports, contracts, and regulatory filings. Built end-to-end ML pipelines in Python and PyTorch for financial risk classification, anomaly detection, and operational forecasting in asset servicing. Integrated Bloomberg and Refinitiv APIs for real-time and historical market data; engineered scalable PySpark pipelines on Azure Databricks with Delta Lake; developed NLP/document intelligence models to extract entities, obligations, and risk indicators; implemented MLOps and LLMOps with Azure ML, Git, Docker, and CI/CD to reduce deployment cycles. Applied Responsible AI practices (SHAP, bias checks) and optimized SQL/Snowflake queries; implemented hybrid cloud workflows to enable legacy system interoperability; built predictive time-series models increasing forecast accuracy by 22%; established monitoring with Azure Mo

AI/ML Engineer at Accenture

Designed supervised learning models (TensorFlow, PyTorch) to predict cardiovascular disease risk from FHIR-compliant EMR data; included feature engineering with Pandas/NumPy for 500k+ records. Developed CNN/RNN architectures (Keras/MONAI) for medical imaging (X-ray, CT) to automate detection of COPD-related anomalies. Built real-time data ingestion pipelines with Kafka and Airflow, processing HL7 clinical events for time-series patient readmission forecasting. Deployed on GCP with Docker/Kubernetes using TensorFlow Serving; 99.9% uptime for clinical decision tools. Conducted bias detection and interpretability analysis with SHAP/LIME; integrated Med-PaLM-inspired LLMs for clinical NLP on unstructured notes, improving diagnostic clarity in ambiguous cases. Optimized PySpark queries on BigQuery for hospital resource planning; automated MLOps with Kubeflow Pipelines and CI/CD; ensured HIPAA/NABH compliance.