University of Southern California — 2026
An internal research tool for assembling cohorts of radiology studies from messy free-text labels — search, filter, and export imaging studies with machine-predicted uncertainty made legible.
A comprehensive system for detecting network anomalies in real-time using machine learning and streaming data processing.
A production-ready ML system detecting fraudulent transactions in real-time using a six-model ensemble with sub-100ms prediction latency.