Based in Paris, France

Hello, I'm
Ziao Wang

PhD Candidate at ENS-PSL.
I work on making physics deep learning, and making deep learning physical, with a focus on free-space optics.

Ziao Wang
Focus
Physics-AI Interface

The Intersection

Bridging Theory & Experiment

I am a researcher working at the convergence of statistical physics, deep learning, and optical computing. My work moves beyond theoretical abstraction to build actual physical systems that learn.

Specifically, I explore how physical principles can constrain and inform deep learning models, and conversely, how deep learning can drive complex free-space optical setups. From deriving statistical mechanics frameworks for neural networks to building custom control boards for optical experiments, I enjoy the full stack of discovery.

Selected Work

Featured Research

Trajectory

Academic Journey

Oct 2023 - Present

Laboratoire Kastler Brossel, ENS-PSL

PhD Candidate

Advised by Prof. Sylvain GIGAN. Building large-scale plug-and-play constrained-ML systems. Leading the "Large Transformer Training with Optics" project.

2021 - 2023

École Normale Supérieure (Ulm)

PhD Track 1st stage (ICFP)

Only 2 seats for the PhD Track. Specialized in Quantum Physics and Statistical Physics.

2016 - 2021

Peking University

B.S. Physics, B.S. Economics

Graduated with Diploma of Weiming (Top 5). Research on Bayesian evaluation.

Technical Expertise

Deep Learning & Code

PyTorch DeepSpeed Transformers Julia C++

Physics & Optics

Free-Space Optics Statistical Physics Replica Method Monte-Carlo

Systems

Raspberry Pi Hardware-in-the-loop Lab Infrastructure

"My approach is rooted in full-stack experimentation. I don't just write the theory; I build the custom control boards to steer the light, design the local storage infrastructure for the group, and write the high-performance code to train the models."

Ziao Wang

Let's Connect

I am always open to discussing physics-constrained deep learning, optical computing, or potential collaborations.