Research
Peer-reviewed and preprint work on AI for health and retail infrastructure. Written to be built on — findings, methods, and, where possible, code and data are open. Full list on Google Scholar.
AI-Enabled Early Detection of Chronic Kidney Disease in Underserved Communities Using Social Determinants of Health
Why it matters: 37M U.S. adults have CKD and most don't know it, with 2–3× higher rates in underserved communities. This models medical plus socio-demographic and SDOH factors to move detection from reactive to proactive.
RetailHealth: Privacy-Preserving Digital Phenotyping for Early Developmental-Risk Estimation
Why it matters: One in six children shows developmental delay, most caught later than optimal. RetailHealth uses household purchasing patterns as an ambient behavioral signal, with an LSTM extracting temporal features to estimate risk — without compromising privacy.
Food-as-Medicine Recommender Systems: A Vision for Generative AI-Powered Grocery Guidance
Why it matters: Reframes the grocery checkout as a point of preventive care — using generative AI to steer everyday food choices toward measurable health outcomes.
Reducing MTTR and Alert Fatigue with PyTorch-Powered Anomaly Detection
Why it matters: The same detection craft, pointed at infrastructure — correlating multivariate telemetry across services and platforms to surface fewer, better alerts with diagnostics and remediation. Presented at the PyTorch Conference 2025 (Linux Foundation).
TODO: Add Intelligent Health Checkout / HSA optimization paper with its DOI/preprint link when ready.