CV
Education
- PhD in Applied Mathematics, DAMTP, University of Cambridge, 2021-2025
- MMath and BA, University of Cambridge, 2017-2021
- (Incomplete) BEng(CompSc), University of Hong Kong, 2015-2017
Work experience
- 2024 Nov-2025 Mar: GSK.ai AI/ML
- Latent geometry of neural networks.
- Supervisors: Dr. Emma Slade, Dr. Alexandra Pettet
- 2023 Jul-Aug: Research Exchange
- University of California, Los Angeles (UCLA)
- New numerical methods for sampling. See this paper.
- Supervisors: Prof. Stanley J. Osher, Prof. Wuchen Li
- 2022 Jan-Jun: External Research Assistant
- Schlumberger
- Numerical method to separate mixed signals with greedy fitting.
- Supervisor: Dr. Can Evren Yarman
- 2021 Aug-Oct: Summer Intern
- Faraday Predictive
- Signal spectral abnormality detection in the low-data regime using clustering techniques.
- Supervisors: Geoff Walker, Andrew Bates
- 2021 Jul-Aug: Summer Intern
- Applied Cryptosystems Department, ASTRI
- Testing federated machine learning frameworks for secure distributed learning.
- Supervisor: Dr. Kam Hong Shum
- 2020 Jul-Sep: Summer Research Assistant
- DAMTP, University of Cambridge
- Researching methods to predict glioblastoma subtypes using MRI and histology.
- Supervisor: Dr. Chao Li
Skills
- Programming
- Main: Python, PyTorch
- Probably OK: JAX, MATLAB
- Very rusty: C, C++, Java, Haskell
- Languages: English, Mandarin Chinese, Cantonese
Publications
Talks
Data-Driven Geometry for Convex Optimization
Group seminar at ETH Zurich Optimization and Decision Intelligence group, Zurich
Blessing of dimensionality using low-dimensional data
PhD symposium talk at GSK.ai, London, UK
Noise-Free Sampling Algorithms via Regularized Wasserstein Proximals Permalink
Mini-symposium Talk at SIAM Conference on Imaging Science 2024, Atlanta, GA
Data-Driven Geometry for Convex Optimization Permalink
Workshop Talk at Big Data Inverse Problems Workshop, Edinburgh, UK
Deterministic Sampling with Wasserstein Proximals Permalink
Poster at Workshop on Optimal Transport: From Theory to Applications, Berlin, Germany
Data-driven Optimization via Mirror Descent
Department Seminar at Oxford-CUHK Joint Seminar Series, Oxford, UK (Online)
Data-Driven Convex Optimization via Mirror Descent
Mini-symposium Talk at ICIAM 2023, Tokyo, Japan
Learned Mirror Descent/Accelerating Plug-and-Play
Department Seminar at Level Set Meeting, Los Angeles, CA
Data-driven Mirror Descent with Acceleration and Robustness
Talk at GSK.ai PhD Symposium, Stevenage, UK
Others
- Third generation scholar of the Masason Foundation
- Reviewer for Inverse Problems and Imaging (IPI), SIAM Journal on Imaging Sciences, ICLR 2025.