Anton Lee
AI research engineer and PhD candidate in AI Engineering at Victoria University of Wellington; my thesis has been submitted for examination. Seeking research and ML engineering roles in New Zealand or Australia, and available to start early 2027. As a New Zealand citizen, I can work in Australia without sponsorship.
My research focuses on continual learning, studying how AI systems can keep learning as data and environments change. In practice, this matters because models degrade as the world evolves and the relationships in the data change. I specialise in detecting this and developing mechanisms to adapt cheaply, whether by improving context or efficiently updating weights (without catastrophic forgetting). Beyond continual learning, I can apply my expertise in machine and deep learning to other AI problems.
Take a look at my thesis abstract.
Selected work
- CapyMOA: co-founded and maintain an open-source Python library for data-stream learning. The CapyMOA paper has been cited 25+ times. Our paper, “Machine Learning for Data Streams with CapyMOA”, appeared at ECML PKDD 2025.
- “Kolmogorov-Arnold Networks Still Catastrophically Forget but Differently from MLP” (AAAI 2025): PDF · code.
- “Look At Me, No Replay! SurpriseNet: Anomaly Detection Inspired Class Incremental Learning” (CIKM 2023): PDF · code.
Skills
Research: Continual learning, data-stream learning, drift detection, anomaly detection, calibration, and dataset shift.
Machine learning: PyTorch, PyTorch Lightning, Hugging Face, scikit-learn; some JAX.
Engineering: Python, Java, C++, Go, CUDA; open-source ML libraries, streaming-method and evaluator development, testing, CI/CD and releases; Linux, Kubernetes, and Slurm.
Service and recognition
Workflow Chair for PAKDD 2027; Publication Chair of the AAAI 2026 Streaming Continual Learning Bridge; tutorials at PAKDD 2025 and PRICAI 2025; AAAI 2025 travel scholarship.
Contact
me@antonlee.kiwi · CV · GitHub · LinkedIn · ORCID