Research
Research
My work sits where estimation theory meets real, messy sensor data. The question running through it is how an autonomous system can stay confident about the state of the world when its measurements are noisy, ambiguous, or missing.
Interests
- Bayesian estimation. Recursive filtering for state estimation in dynamic, uncertain environments.
- Data association. Deciding which measurement belongs to which object when there are many of both.
- Time series classification. Machine learning that detects and diagnoses faults in aerospace systems from telemetry.
Current work
Detection & Tracking for mWidar Imaging
Detecting and tracking moving targets in continuous-capture microwave imagery from mWidar, a low-cost radar that images a whole scene at once at thousands of frames per second.
Time Series Classification for Satellite Fault Detection
Teaching spacecraft to diagnose their own faults (power, attitude control, propulsion) directly from telemetry, without waiting on a ground operator.
Past work
Spectral-ViT: Transformers for Supernova Spectra
My undergraduate senior thesis: Spectral-ViT, a transformer model I developed to classify supernova spectra from the DESI survey.
Rotational Dynamics of Binary Asteroids
How non-principal axis rotation in binary asteroid systems weakens the BYORP effect. Published in Icarus.