At MedStar Health, I build machine learning and deep learning analyses with physicians and faculty investigators to answer clinical research questions. I develop cohorts and model evaluation workflows using Python, SQL, and Databricks.
For the RECOVER Alexa Study, I build production pipelines that integrate EHR, Alexa and chat interactions, and REDCap patient-reported data. I also develop Power BI dashboards for clinical teams and NIH-facing reporting.
For RECOVER Clinical AI, I evaluate a production pre-trained LLM for patient symptom conversations through prompt and response testing, clinical appropriateness review, and edge-case analysis. I design response-quality and escalation checks with human-in-the-loop safety review.
Earlier, at Johns Hopkins University’s Center for Digital Health and AI, I built analytics pipelines across 1.57M records and 15,000+ users, and contributed to peer-reviewed EHR-based risk stratification research. My work also includes data integration and validation for cancer research at MedStar Health, and analytics experience at NetEase / ByteDance.
