University of Miami
Research Associate II
I was brought on to assume multiple responsibilities across several projects. My biggest accomplishments are centered around turning raw motion and audio data into reliable signals of developmental delays in children. I led a ground-up rebuild of production ML models and created new features, drawing from my background in physics, which improved performance on hard-to-detect cases of ASD. I also led the development of an on-device embedding model using small language models to analyze conversational turn-taking. In parallel, I helped build an automated transcripted tool that has been adopted across 7 preschools and used internally in the lab, and I contributed to a computer vision system that allows non-technical users to analyze videos using natural language. My work has aided the workflow of technical and non-technical researchers and collaborators, and has been part of the foundation of other studies. Along the way, I have contributed to manuscripts that are currently under review at INSAR, ICDL, and Advanced Sciences.
Miami Country Day School
Physics Teacher
I taught high school physics at all levels: college prep, honors, and AP. I take pride in having created an experiment-first curriculum that helped bridge a gap between science and students' interests. When schools shut down as a response to the pandemic, I developed a graphing tool from scratch that students could use to perform experiments at home and analyze their data. Even during a difficult year, my students had higher average test scores at the local, state, and federal level.
Below are some groovy side projects
Tod Talk Demo - Transcript Analysis Tool
Tech stack: Python, NexaSDK, SQLite
Workout Rep Counter
Tech stack: Swift, MediaPipe
Basketball Tracker
Tech stack: Python, PyTorch, OpenCV
