Led the development and implementation of a clinically-informed metric for evaluating an internal entity linking model, leading to improved model performance and enhanced collaboration across teams.

Trained medical AI image segmentation models using PyTorch to isolate kidneys and ureters on CT scans and grade lumbar stenosis on MRI scans. Partnered with GE Healthcare and MGH to ship models.

Designed and trained a graph-based AI model using TensorFlow to generate embeddings for biomedical entities and relations in the system’s knowledge graph. Worked closely with the CEO to lay the foundation for its integration into the product’s retrieval service.

Published a first-author paper in American Medical Informatics Association (AMIA) Informatics Summit benchmarking AI de-identification techniques for identifying HIPAA-regulated tokens in Electronic Health Records (EHRs).

Served as graduate teaching assistant for CSCI 1470: Deep Learning and BIOL 1595: AI in Biomedicine.

Published several first-author papers in the health AI space, including in delirium detection, precision oncology, and COVID-19 prognostication.

Wetlab research on aging using genetic assays and molecular techniques for studying Drosophila flies.