Machine learning for the human heart.

About me

I grew up in Portland, Oregon, and started doing research in high school. I've spent over six years using machine learning for early detection and prevention of cardiovascular disease — starting with simple ECG classifiers, and now focused on synthesizing full 12-lead ECGs for at-home monitoring. My research has taken me to international conferences around the world: first-authored papers at IEEE EMBC and IEEE SENSORS, presentations at the Heart Rhythm Society and AI Med, and two International Science and Engineering Fair (ISEF) Grand Awards.

I currently study computer science at Carnegie Mellon. Along the way I've also captained a FIRST Tech Challenge robotics team for four years and run my small baking business, TnT Treats, since 2019. In 2026 I was named a Coca-Cola Scholar — one of 150 selected from over 110,000 applicants.

Research & projects

I work on developing AI algorithms for the heart, used for early detection, diagnosis, and monitoring of cardiovascular disease. Over the years I've developed a range of algorithms toward those goals.

My current main focus is ECG synthesis. A clinical ECG needs twelve leads and ten electrodes; my work synthesizes a diagnostic-quality 12-lead ECG from a subset of leads that a low-cost wearable can record. That makes at-home monitoring and early detection of cardiovascular disease easier and more accessible. It also lets low-resource regions get high-quality ECGs without the need for expensive ECG recording devices.

Outside this line of research, I've also built IntelliCane, an autonomously navigating cane that actively guides the visually impaired.

1. 2025
   Transformer-Based Framework with Dynamic Multiscale Attention Mechanisms for 12-Lead ECG Synthesis

   47th IEEE EMBC · First-author full paper · IEEE Xplore & PubMed

   I first-authored and presented this paper at the world's largest biomedical engineering conference. I created a custom transformer that uses dynamic scaling windows to capture both the sharp spikes and the subtle waves of a heartbeat, allowing for accurate ECG synthesis. My model can also be fitted to accurately synthesize any type of parallel waveform with minimal work.

   Project page → Paper (PDF) Slides (PDF)
2. 2025
   A Morphologically-Aware Transformer Network for High-Fidelity V1–V5 ECG Lead Generation

   24th IEEE Sensors · First-author full paper · IEEE Xplore

   I first-authored and presented this paper at IEEE's flagship sensor conference. Where my EMBC model is general-purpose, this multi-stage transformer is customized to ECGs: the architecture builds in the specific features and complexities of heart waveforms with each stage focused on synthesizing different parts of the ECG (P, QRS, and T segments).

   Project page → Paper (PDF) Poster (PDF)
3. 2025
   IntelliCane: A Deep Learning Cane for Active Obstacle Avoidance & Navigation for the Visually Impaired

   Regeneron ISEF 2025 · Grand Award, 3rd place · AAAI Special Award

   I built the first goal-driven, self-navigating aid for the visually impaired. IntelliCane can interpret destinations in plain language — like “go to a water fountain” — and its camera-based AI agents detect obstacles, find a walkable path, and actively guide the user there with servo steering and spoken directions.

   Project page → Abstract (PDF) Poster (PDF)
4. 2024
   Validation of Einthoven and Goldberger's Electrocardiographic Lead Systems

   Heart Rhythm Society 2024 · Abstract & poster

   I re-validated the century-old formulas that modern electrocardiography is built on, comparing their computed leads against real recordings across a range of heart conditions. I presented the work at the Heart Rhythm Society's annual meeting, showing how essential these formulas are for detecting more types of cardiovascular disease from a wearable device with a limited lead set.

   Project page → Abstract (PDF) Poster (PDF)
5. 2024
   Synthesis of a Diagnostic-Caliber 12-Lead ECG Using a Novel Multi-resolution Transformer

   Regeneron ISEF 2024 · HRX 2024 · Grand Award, 2nd place

   I built my first end-to-end transformer pipeline for generating a diagnostic-quality 12-lead ECG from three recordable leads, pairing the mathematical lead-computation formulas with a multi-resolution neural network personalized to each individual. This is the project that launched my research, and I later presented it at HRX 2024.

   Project page → Abstract (PDF) Poster (PDF)
6. 2023
   Early Detection and Synthesis of a 12-Lead ECG Using Density-Based Clustering and GANs

   AI Med 2023 · Youngest of 80+ presenters

   I trained a GAN to generate a 12-lead ECG from the three leads a smart device can record, then used clustering against a person's own ECG history to flag early signs of heart disease without the need for pre-labeled data. Because the baseline is the person's own heart, it works even for people with pre-existing conditions, where standard classifiers fall short. I presented this work at AI Med 2023 as the youngest of 80+ pre- and post-doctoral presenters.

   Project page → Slides (PDF) Abstract (PDF)

Awards

1. 2026
   Coca-Cola Scholar. $20,000 national merit scholarship; 150 recipients from more than 110,000 applicants.
2. 2025
   Regeneron ISEF Grand Award, 3rd place, Robotics and Intelligent Machines, plus the AAAI Special Award.
3. 2024
   Regeneron ISEF Grand Award, 2nd place, Robotics and Intelligent Machines.
4. 2023
   AI-Med Youngest Presenter, among a field of 80+ pre- and post-doctoral researchers.

Skills

Contact

I'm always up for collaborations or talking more about my research. Email is the fastest way to reach me.