I am a computational scientist interested in generalizable machine learning. I am pursuing double bachelors degrees in Mathematics and Linguistics at Yale University, where I am a member of Saybrook College (the best residential college). You can download a copy of my CV here. I am a member of the CLAY research group where I work on the following topics:
How do neural networks generalize: How can we design neural networks which perform well on data which is outside their training domain? What kinds of inductive structural biases exist in current neural architectures?
How do neural networks work: How are neural networks able to make the kinds of inferences we observe? What are the bounds on the computational expressiveness of neural architectures? What do pre-trained language models know about the languages they are trained on?
What can ML tell us about human learning: Can we create ML models which exhibit the same learning patterns as children during language acquisition?
Outside of research, I’ve interned at a variety of different organizations, including:
Novetta, where I currently focus on designing multimodal perception models to perform real-time inference on simultaneous feeds of different kinds of input data.
HELIX re, where I built models to perform semantic segmentation on LiDAR scans of buildings to enable rapid development of digital building models.
The US Forest Service, where I helped to develop tools to model the progression of forest fires in western Montana, allowing for better predictions and resource allocation during fire season.
Questions or comments about research may be addressed to
Elsewhere on the internet
This site is made using Pollen, a publishing system developed in Racket by Matthew Butterick. Text is set in Production Type’s Newsreader and Huerta Tipográfica’s Alegreya Sans. Monospace text is set in Menlo.