Research.

Real-time control of tokamaks — foundation models, computer vision, and signal processing for plasma state estimation and instability prediction. Advised by Egemen Kolemen at Princeton MAE / PPPL.

Research

TokEye

Find the structure hiding in a noisy signal.

A tool for extracting coherent and transient features from fluctuating time series—turning a crowded spectrogram into something a researcher can measure.

OracleFlow

How much better could a model still get?

A controlled setting for studying learning: separate ambiguity in the data from the error a model could still reduce.

Fusion World Model

Learn how a plasma evolves. Help shape what happens next.

A world model for fusion, built from scratch to connect diagnostic measurements, learn plasma dynamics, and support better experimental decisions.

DIII-D control

Turn a camera measurement into a control decision.

Image-based feedback for divertor detachment at DIII-D: connecting what the camera sees to how an experiment responds.

Fig. 01 · TokEye — fast signal extraction for fluctuating tokamak time series.

Publications

publications: 13
citations: 9
institutions: 9

Towards Large-Scale Heterogeneous Data Organization for Scientific Foundation Models: A Nuclear Fusion Case Study

arXiv preprint arXiv:2608.275782026

Robust control of ECH deposition profiles on DIII-D

arXiv preprint arXiv:2606.136612026cited by 1

Compact Experimental Negative TriAngUlarity Reactor (CENTAUR): Design Study for a Breakeven Negative Triangularity Device

67th Annual Meeting of the APS Division of Plasma Physics2026[preprint]cited by 1

Regulation compliant AI for fusion: explainable image-based feedback control of divertor detachment in DIII-D tokamak

Nuclear Fusion 66 (3), 0360142026cited by 6

TokEye: Fast Signal Extraction for Fluctuating Time Series via Offline Self-Supervised Learning From Fusion Diagnostics to Bioacoustics

arXiv preprint arXiv:2602.203172026

Beyond the Loss Curve: Scaling Laws, Active Learning, and the Limits of Learning from Exact Posteriors

arXiv preprint arXiv:2602.003152026

Real-time feedback control of radiation front-based detachment enabled by machine learning on DIII-D and KSTAR

67th Annual Meeting of the APS Division of Plasma Physics2025cited by 1

Real-Time Machine-Learning Enabled Emission Front Control at DIII-D

Bulletin of the American Physical Society2024

Visual Perception of 3D Space and Shape in Time-Part IV-3D Shape Recognition by 3D Rotation

bioRxiv, 2022.03. 01.4821642022

What Lies Beneath the Curve? Scaling Laws in the Presence of Exact Posteriors

ICLR 2026 2nd Workshop on Deep Generative Model in Machine Learning: Theory …

Dynamic modeling of Alfvén eigenmodes using Machine Learning on DIII-D

67th Annual Meeting of the APS Division of Plasma Physics

Self-Supervised Identification of Coherent Modes in Tokamaks

67th Annual Meeting of the APS Division of Plasma Physics

Towards a Foundation Model for Fusion: Multimodal Representation Learning of Plasma State and Control

67th Annual Meeting of the APS Division of Plasma Physics

Synced from Google Scholar · Sep 8, 2026