Machine Learning Research

During my research with the LAIR Lab at the University of Colorado Boulder, I developed a pipeline to analyze satellite data of energetic particle precipitation. ELFIN satellite data was converted into standardized time–energy plots, where different precipitation mechanisms produce distinctive visual signatures (e.g., “blobs” or "wave-curl" structures).

I built software to efficiently generate these plots at scale, constructed a labeled dataset of events, and trained custom YOLOv8 computer vision models to classify precipitation sources. The trained models were then applied across more than six years of multidimensional satellite observations to automatically identify candidate events. This work enabled large-scale analysis of space weather activity and was presented at the National Conference for Undergraduate Research.