At Novant Health, I build RAG pipelines that generate discharge and follow-up summaries from clinical notes and EHR data, reducing care-team follow-up response time by approximately 34%.
I also develop patient-readmission risk models combining structured EHR data with transformer embeddings, improving prediction accuracy by approximately 27%. I build retraining workflows, distributed data pipelines, and clinician-facing explainability dashboards while working with HIPAA-compliant clinical data.
Previously at HCLTech, I built real-time fraud-scoring, KYC automation, and production MLOps systems for high-volume payment platforms. My work reduced false positives by 31% and model-drift-related incidents by 25%.
