• I was a Lead Data Scientist at C3.ai, where I built enterprise AI systems with a main focus on LLM pretraining, finetuning, and autonomous agentic systems.

• I obtained my PhD from The Hong Kong Polytechnic University, Hong Kong under the supervision of Prof. Kainam Thomas Wang. From 2014 - 2016, I was a Postdoc at University Paderborn, Germany, working with Prof. Peter Schreier. From 2016 to 2022, I have been working as a Senior Research Fellow with Prof. Tay Wee Peng at Nanyang Technological University, Singapore.

• My research interests include robust machine learning, statistical signal processing, LLM finetuning, and autonomous agentic systems.

Recent News

• Our paper entitled “NEMO: Execution-Aware Optimization Modeling via Autonomous Coding Agents” has been accepted by ICML 2026 (acceptance rate: 26.6%).
• Our paper entitled “Unleashing the Potential of Fractional Calculus in Graph Neural Networks with FROND” has been accepted by ICLR 2024 as Spotlight (acceptance rate: 31%).
• Three papers have been accepted by AAAI 2024 (acceptance rate: 23.75%): PosDiffNet, DistilVPR, and Coupling GNN with Fractional Order Dynamics.
• Our paper entitled “Adversarial Robustness in Graph Neural Networks: A Hamiltonian Approach” has been accepted by NeurIPS 2023 as Spotlight (acceptance rate: 26.1%).
• Our paper entitled “Node embedding from neural Hamiltonian orbits in graph neural networks” has been accepted by ICML 2023 (acceptance rate: 27.9%).
• Our paper entitled “Graph neural convection-diffusion with heterophily” has been accepted by IJCAI 2023 (acceptance rate: 15%).
• Our paper entitled “HypLiLoc: Towards effective LiDAR pose regression with hyperbolic fusion” has been accepted by CVPR 2023 (acceptance rate: 25.8%).
• Our paper entitled “On the robustness of graph neural diffusion to topology perturbations” has been accepted by NeurIPS 2022 (acceptance rate: 25.6%).
• One paper entitled “Stable neural ODE with Lyapunov-stable equilibrium points for defending against adversarial attacks” has been accepted by NeurIPS 2021 (acceptance rate: 26%). Code
• From July 2022, I have started a new adventure in industry.
• One paper entitled “Error-correcting output codes with ensemble diversity for robust learning in neural networks” has been accepted by AAAI 2021 (acceptance rate: 21%).