PoHan (Kozak) Hou
CURRENTLY PhD Student in Nuclear Fusion Research Program
Department of Complexity Science and Engineering
University of Tokyo
ALSO Scientific AI & HPC Consultant
Full Stack Software Engineer (FDE), XingAn Clinic
PREVIOUSLY MSc Applied Computational Science and Engineering
Imperial College London, United Kingdom BS Space Science and Engineering
National Central University, Taiwan

PoHan (Kozak) Hou

FUSION × MACHINE LEARNING × HPC

I research Physics-Informed Neural Operators for fusion plasma at the Ejiri - Tsujii Laboratory in University of Tokyo , where I'm accelerating two-dimensional lower hybrid wave simulation for the TST-2 spherical tokamak. Alongside my PhD, I architect ML systems in industry and consult for teams who need scientific simulation to run faster.

My work sits at the intersection of scientific machine learning, high-performance computing, and applied physics.

RESEARCH FOCUS
Reg-PINNs, PINNs & neural operator methods
Geometry-embedded Fourier Neural Operators
Lower hybrid wave & plasma simulation, TST-2
HPC & scalable scientific workflows
SCIENTIFIC AI & HPC CONSULTING
Helping engineering teams accelerate scientific simulations with AI.
CFD acceleration · Neural surrogates · plasma & electromagnetic simulation · heat transfer · multiphysics workflows
Start a conversation →