My research interests lie in the intersection of explainable/interpretable AI, computer vision, and medical imaging. I find myself first interested by trying to understand how AI is producing the answers it does. This has broader societal and ethical implications relating to trust, which is why the use and audit of it in the medical field is of particular interest.
For my M.S. thesis, I'm extending my previous work using image translation networks for classification by examining the interal activations of the network and adding uncertainty quantification.
I'm a current M.S. student in Computer Science at Colorado State University, advised by Jesse Wilson and Ravi Mangal. My research focus in the last several years has shifted to the idea of explainable AI and uncertainty quantification.
This work was supported by a Boettcher Collaboration Grant as a partnership between Jesse Wilson, a Boettcher Investigator, and me, a 2020 Boettcher Scholar.
Journal of Medical Imaging 13(S1), S11207. Journal article.
Medical Imaging 2026: Computer-Aided Diagnosis, Proc. SPIE 13926, 139260F, pp. 99–107. Conference paper, presented as a talk in Vancouver.
A local-first experiment tracker for ML workflows, published on PyPI. It records parameters, metrics, git state and code changes from training scripts and notebooks into a single SQLite file, using only the Python standard library.
pip install exptrack
Build a neural network in the browser and step through it one layer at a time, with the real tensor values, the equation, the gradient and the matching PyTorch code at each step.
Drop in a folder or paste a GitHub URL and get an interactive map of an unfamiliar codebase. It runs entirely in the browser, with no server and no LLM.