About Me
Bio
I am a PhD candidate at the Applied Artificial Intelligence Institute (A²I²), Deakin University (Australia), supervised by Prof. Truyen Tran and Dr. Hung Le. My research focuses on Sparse Mixture-of-Experts (MoE) for efficient training and inference of large language models — studying routing, representational collapse, and the emergence of specialised experts — and on controllable and robust reasoning in LLMs. A unifying thread of my work is the binding problem: how foundation models compose, separate, and route information across modular components.
I hold an M.Sc. in Data Science from VNU University of Science (Hanoi, Vietnam). I also hold a B.Tech. in Information Technology from Hanoi University of Science & Technology and a B.A. in International Business Economics from Foreign Trade University.
Before academia, I spent more than six years as a data scientist and AI researcher, leading and contributing to industrial projects at Panasonic R&D Vietnam, Techcombank, and Toyota Motor Vietnam.
Research Interests
- Sparse Mixture-of-Experts (MoE): routing, representational collapse, specialisation.
- Efficient training and inference of large language models.
- Controllable, robust, and interpretable reasoning in LLMs.
- Foundation models and the binding problem.
News
- 2026 — Two papers accepted to ICML 2026 (one Spotlight).
- 2026 — One paper accepted to ACL 2026 (Findings).
- 2025 — SimSMoE accepted to NAACL 2025 (Findings); On the Role of Discrete Representation in Sparse MoE accepted to TMLR 2025.
- 2024 — Joined the Applied AI Institute (A²I²), Deakin University, as a PhD candidate.
- 2024 — Awarded UTC Outstanding Graduate Student, 2023–2024.
- 2023 — HyperRouter accepted to EMNLP 2023 (Main Conference).