Biomedical Machine Learning Scientist
Developing Next-Generation AI to Predict Phenotype from Genotypes.
Department of Biomedical Informatics
Harvard Medical School
Boston, MA 02115
Email: yasha_ektefaie@g.harvard.edu
Google Scholar
Who am I?
I am a PhD candidate in the Bioinformatics and Integrative Genomics program at Harvard Medical School, and a National Defense Science and Engineering Graduate (NDSEG) fellow, co-advised by Dr. Maha Farhat and Dr. Marinka Zitnik. I am interested in creating AI that can (1) understand whole genome sequencing data to predict phenotypes ranging from antibiotic resistance to Tuberculosis to single-cell gene expression and (2) generalize to unseen sequences. Previously, I have designed convolutional neural networks to predict breast cancer presence and subtypes from histopathology slides with Dr. Kun-Hsing Yu. I received a B.S. in Electrical Engineering and Computer Science (EECS) and another BS in BioEngineering (BioE) from UC Berkeley, where I developed computational methods to understand microbial communities with Dr. Adam Arkin and Dr. Lauren Lui. Beyond academia, I have worked at Flagship Pioneering creating a novel protein inverse folding model, at Dascena (now CirrusDx, Inc) designing machine learning models to use EHR data to predict ischemic stroke occurrence in patients, at Verily gathering metrics for a convolutional neural network predicting diabetic retinopathy to facilitate FDA consideration of the algorithm.