projects
papers
-
Panda: Self-distillation of Reusable Sensor-level Representations for High Energy Physics
We apply DINO-like self-distillation to unlabeled 3D images of particle interactions in a liquid argon time projection chamber. We show the model learns to cleanly separate particles and interaction types in latent space, which reduced the number of labeled images required to reach the state-of-the-art for semantic segmentation by 1,000x.arXiv, 2025 -
10 Million Particle Events: Enabling Foundation Models for Sparse 3D Inverse Problems
A proposal for a new dataset containing 10 million images of particle interactions with realistic detector response and multiple modalities.NeurIPS 2025 AI4Science Workshop (Spotlight), 2025 -
Particle trajectory representation learning with masked point modeling
We proposed a masked point modeling approach for learning representations of particle trajectories from unlabeled LArTPC images. PoLAr-MAE learns to separate tracks from showers almost perfectly without any labels.Machine Learning: Science and Technology, 2025
news
- I was awarded the HAI Graduate Fellowship.
- Won Stanford’s CS 229 Machine Learning’s Best Project Award for my rotation work on pileup synthesis and anomaly detection for the ATLAS experiment.
- I graduated from Penn with a bachelor’s and master’s in physics, and will continue my studies at Stanford.
- I’m extremely grateful to receive the Roy and Diana Vagelos Challenge Award (two years full tuition and fees) at Penn.
talks
- CHEP, “Toward a Foundation Model for Neutrino Physics: Self-distillation of Reusable Sensor-level Representations” Slides
- HAI+SDS Annual Conference, “Learning the Structure of Particle Interactions From Raw Detector Data Without Labels” (Lightning Talk)
- NPML, “Toward a general-purpose foundation model for neutrino physics” Slides
- ML4FP Summer School, “Toward a general-purpose foundation model for neutrino physics” Slides
- APS Global Summit, “A foundation model for LArTPC events”
- APS April Meeting, “Differentiable surrogate for modeling the physics of optical propagation in a LArTPC”
- 16th Marcel Grossmann Meeting, “The Optical Two- and Three-Dimensional Fundamental Plane Correlations for More than 130 Gamma-Ray Burst Afterglows”
- APS April Meeting, “Impact of Spectral Photon Sorting in Large-Scale Neutrino Detectors”
- APS Mid-Atlantic Section Annual Meeting, “Impact of Spectral Photon Sorting in Large-Scale Neutrino Detectors”
misc
chroma-cam
Visualize neutrino detectors in the browser with meshes
PoLArMAE Attention Viz
Visualize the learned attention maps from PoLArMAE
visualizations
miscellaneous visuals and tools I've made over the years