PHAST: Port-Hamiltonian Architecture for Structured Temporal Dynamics Forecasting
arXiv preprint, 2026
PhD Student at UT Austin · Co-founder, Mihawk.ai
I build safe, efficient, and robust learning agents. My research spans dynamical systems, scientific ML, 3D vision, and generative modeling at the Computational Visualization Center, advised by Prof. Chandrajit Bajaj.
My work focuses on building learning systems grounded in physical principles. Current interests include:
I am also a professional yoga trainer, exploring how yoga can optimize:
"I strive for a calm mind, fit body, and a heart full of love."
arXiv preprint, 2026
bioRxiv, 2025
ICONIP 2024 · arXiv:2410.15208
For earlier work and industry publications, see Google Scholar and Patents.
Reach me at shubham.bhardwaj [at] utexas [dot] edu or connect on LinkedIn.