Achieved an AUCPR of 0.67 four hours before onset time, where previous research reached only 0.4

Automated patient labeling, imputed missing patient data with gaussian processes, and engineered over 600 time series features

Created a reproducible data pipeline to query UCSF database using Spark and Python

Modeling using Random Forest, Gradient Boosting, and RNN’s using PyTorch

Evaluated assignments and exams for Introduction to Computer Science