cv
General Information
| Full Name | William (Chi Kin) Yau |
| yau5@purdue.edu | |
| Phone | (341) 766-9170 |
| Website | williamcyau.github.io |
| linkedin.com/in/williamcyau | |
| Location | West Lafayette, Indiana, US |
Education
-
Aug 2025 - Now Ph.D in Electrical and Computer Engineering
Purdue University GPA 4.00/4.00- EUV & comp. lithography (Abbe/Hopkins, OPC/ILT, SMO, M3D)
- nanophotonics & metamaterials
-
2021 - 2025 BA in Physics and Computer Science
University of California, Berkeley GPA 3.93/4.00- image processing & vision (C++), computational geometry
- parallel computing (C, Linux)
- laser & optoelectronics (Lumerical), Fourier optics (Zemax)
- physics-informed machine learning, data analytics
Research Experience
-
Aug 2025 - Now Graduate Researcher
Purdue ECE Stanley Chan Lab Developing physics-aware, learning-based OPC framework to optimize masks with interpretable procedures.- Training GPU-accelerated Fourier neural operator-based surrogate models of Hopkins imaging for generating extensive simulated datasets of mask/aerial/resist images, ensuring model robustness and generalizability.
- Developing variational autoencoder (VAE)-based generative image encoding for aerial images with PyTorch, improving gradient-based optimization landscape, offering well-posed solutions and algorithmic interpretability.
- Deriving theoretical guarantees on optimization landscape properties (benign non-convexity), ensuring convergence speedups and mathematical implications on illumination design and SMO.
- Benchmarking framework against OPC metrics (MEEF, NILS, PV band) to provide interpretable knobs for imaging product engineers and customers.
-
Aug 2025 - Now Graduate Researcher
Purdue ECE Qi Guo Lab Developing an all-optical framework for algorithm-metasurface co-design based on EM simulations of wafer stacks to quantify sub-micron line/space surface defects with electron-microscopy-level precision.- Achieved 5-10X improvement in defect location pinpointing and 1.3-3X higher defect-severity accuracy (<10 nm error) across large 3D patterned fields using end-to-end EM optimization.
- Developed RCWA-based figure of merit for PyTorch-based co-optimization of scatterometric illumination/detection and metasurface design parameters, ensuring optimal imaging accuracy and noise-robustness.
- Deployed GPU-accelerated FDTD-based EM solver for 3D volumetric fields around multilayer stacks with diverse realistic defects (CD non-uniformity, sidewall roughness) for generalizable data-driven imaging.
- Implemented high-dimension parameter sweeps over defect configurations (CD, sidewall, height) across multi-GPU clusters, leveraging automated job scheduling, MPI parallelization, and GPU-direct data pipelines in Python/CUDA for 10X faster EM simulation throughput on realistic wafer geometries.
-
Aug 2024 - 2025 Research Assistant
Berkeley Artificial Intelligence Research (BAIR) Laura Waller Lab Modeled aberration wavefront profiles with Zernike polynomials to study their impact on imaging contrast and resolution in microscopy/astronomy systems, connecting with a proprietary information-theoretic, data-driven framework for holistic, generalizable object-aware optical system characterization.- Built scalable Bayesian model pipelines with GPU acceleration for image acquisition modeling & validation.
- Integrated JAX-TensorFlow-PyTorch modules for differentiable inverse design pipelines for diffractive optical elements (akin to lens/illumination design optimization), improving accuracy 25-30%.
-
May - Jul 2024 Research Fellow
Purdue ECE Stanley Chan Lab Developed signal-denoising and depth-estimation algorithms boosting SP-LiDAR depth accuracy 1000X. Developed signal-denoising and depth-estimation algorithms boosting single-photon LiDAR (SP-LiDAR) depth estimation accuracy 1000X and noise robustness 17X, enabling rapid, on-chip processing, addressing data storage limitations common for SP-LiDAR; Wrote unit tests and regression tests in C++/Python simulation modules to ensure numerical stability and performance.- Wrote unit tests and regression tests in C++/Python simulation modules.
-
Aug 2022 - 2023 Research Assistant
UC Berkeley Space Sciences Lab (SSL) Ivan Vasko Group Developed scalable Python/Pandas-based data pipeline & dashboard processing 1M+ plasma dynamics datapoints from NASA THEMIS, enabling high-throughput analytics and performance monitoring.
Projects
-
Dec 2025 RCWA-based Simulation of Overlay Metrology using OCD (Scatterometry)
RCWA solver in Python/CUDA modeling scatterometry-based overlay metrology (ASML YieldStar-like). Distributed parameter sweeps on multi-GPU Linux clusters for configuration optimization. -
Dec 2025 RCWA-based Surface Defect Field Simulation
GPU-accelerated RCWA solver generating image libraries for line/space patterning defects. Compared vs. FDTD; analysis scripts computing robustness and image quality metrics. -
Dec 2025 Plane Wave Expansion Method (PWEM)-based Isolated Nanopillar Field Sim.
GPU-accelerated PWEM eigen-solver for arbitrarily shaped nanopillars. Parameter sweeps and data preparation pipeline for ML-based surrogate modeling.
Awards and Presentations
-
2021-2025 Hong Kong Scholarship For Excellence Scheme (HKSES) Awardee
Four years of full-tuition at UC Berkeley. -
Apr 2025 Hong Kong Scholarship For Excellence Scheme (HKSES) Awardee
Physics-inspired Neural Mapping for High-flux Single-photon LiDAR -
Oct 2024 IEEE MMSP Conference
Analysis and Improvement of Rank-Ordered Mean Algorithm in Single-Photon LiDAR
Technical Skills
-
Languages
- Python, C++, C, MATLAB, Java, Git
-
Image Processing
- OpenCV, TorchVision, GEOS, Dlib, Simd
-
Data/ML
- PyTorch, JAX, TensorFlow, Pandas, NumPy, SciPy
-
HPC
- CUDA, MPI (Linux), Dask
-
EM Simulation
- Meep, Tidy3D, JCMsuite, TorchLitho, OpenILT
-
Ansys EDA
- Lumerical, Zemax, Code V, APDL
Volunteering/Community Work
-
2024 - 2025 Editorial Photographer
Fashion and Student Trends @ Cal Shot and directed creative campaigns, including clients like Valentino Garavani. -
2022 - 2025 Senior Photographer
The Daily Californian Work frequently featured on website and printed copies' front page. -
2021 - 2022 Tech & Operations Associate
UC Berkeley ASUC Built AWS infrastructure and learning-based algorithms for Connect@Cal resource distribution.