COVID-19 FORECASTING PIPELINE
Delivered a data-driven compartmental modeling pipeline to produce timely forecasts for COVID-19 deaths and cases across all countries and many subnational locations. The results were used for policy decisions and pandemic planning, and were widely featured in national news media, including The New York Times, The Washington Post, CNN and NPR.
Developed new epidemiological models to estimate complex relationships for a wide range of risks and diseases. Implemented a robust estimation pipeline to synthesize data across studies using in-depth understanding of data reporting mechanisms.
Developed the core model for estimating anti-microbial resistance from sparse and noisy datasets. Turned expert knowledge into modeling constraints to complement limited data.
Invented a flexible, robust and stable approach to estimate the efficiency frontier for healthcare availability as a function of cost of services.
AIRLINE ANOMALY DETECTION
Automated a manual anomaly detection process for an airline company using a combination of time series analysis and new machine learning methods.
UNDERWATER TRACKING AND NAVIGATION
Developed and implemented robust filtering and smoothing algorithms to track seaglider vehicles for a government research laboratory.
Provided statistical analysis of multiple datasets to determine biomarkers for predicting hyperbilirubinemia in newborns.
Analysis Plan for Pharmaceutical Study
Developed a statistical analysis plan for a medical research grant that estimated post-partum dynamics of biomarkers.