Robert Sandor

Software Engineer

University of California, San Francisco•  November 2018 - Present

• Created preprocessing pipeline and trained a U-Net Convolutional Neural Network (CNN) in PyTorch to help triage patients after brain surgery. Segmented brain MRI scans 3% better than existing software with 14% less variance. • Applied transfer learning from MRI scan-based model to CT scan-based model to obtain similar performance using the FastAI library • Implemented Continuous Bag of Words (CBOW), Long Short Term Memory (LSTM) and CNN Natural Language Processing (NLP) models to classify medical notes.

University of San Francisco

Data Science, MS•  July 2018 - Present

Coursework: Deep Learning, Machine Learning, Distributed Computing, Experiments in Data Science, Computational Statistics, Time Series Analysis, SQL, Data Visualization, Linear Regression