Haoyu Dong

I am currently a Statistics undergraduate at the University of Glasgow. My recent work combines interaction design and machine learning to build intelligent interactive systems that support multimodal intent expression in scientific workflows. I have also explored geometric approaches to graph representation learning.

My primary interests lie in human–AI interaction, LLM-assisted interfaces, and intelligent interactive systems. I am curious about how people communicate their ideas and intentions to AI, understand its responses, and stay involved as their ideas evolve. I would also like to explore interactive AI agents, as well as how wearable computing, extended reality (XR), and spatial interaction can support these experiences beyond the desktop.

Ultimately, I want to build systems that people can use in their everyday lives, taking research ideas beyond prototypes and turning them into tools that make a meaningful difference in how people interact with the world :)

I am currently seeking PhD opportunities in HCI, human–AI interaction, and related areas. If you think my interests might be a good fit for your group, I’d love to hear from you!

Selected publication

UIST 2026

MolecularCanvas: LLM-assisted Small-Molecule Drug Discovery via Structure-Guided Constraints

Haoyu Dong*, Rui Sheng*, Shuhao Zhang, Yushi Sun, Dingyang Wu, Hanxiang Chao, Olexandr Isayev, Huamin Qu, Yuyang Wu, and Yanna Lin

* Equal contribution

Paper PresentationYouTube ↗
Conference DemoYouTube ↗

Want to try it yourself?

Have a play with MolecularCanvas—edit a molecule and see how the pieces fit together.

UI demo · simulated results
Try it out

Research experience

2025–2026

Oct 2025 – Jul 2026

MolecularCanvas: LLM-assisted Small-Molecule Drug Discovery via Structure-Guided Constraints

Lead Researcher · HKUST VisLab · Mentored by Dr. Yanna Lin

Led the interaction and system design of a human–AI molecular optimisation platform. Built the full-stack prototype and led expert studies with medicinal and computational chemists.

Mar 2026 – Jul 2026

GRAPPLE: Curvature-Adaptive Grouped Geodesic Prototype Learning for Node Classification

Lead Researcher · Advised by Prof. Xuchu Jiang

Developed a plug-in GNN classifier with grouped prototypes in a learnable constant-curvature space, leading the method, implementation, experiments, and manuscript. Under review at ACM KDD 2027.

Feb 2025 – Aug 2025

DECAGON: Decaying Self-Paced Graph Structure Learning with Global Alignment

Lead Researcher · Advised by Prof. Xuchu Jiang

Proposed a self-paced graph structure learning method with decaying masks, error-memory, and Gromov–Wasserstein global alignment. Manuscript under revision.

Education

2025–2027 expected

University of Glasgow

BSc (Hons) Statistics · Dual-Degree 2+2 Programme

Year 3 GPA: 18.8/22.0 · First-class level

2023–2025

Zhongnan University of Economics and Law

BSc Statistics · Dual-Degree 2+2 Programme

Weighted average: 85/100

Teaching

2026–Present

Statistics Tutor and Demonstrator

University of Glasgow · Level 1 and Level 2 Statistics

Honors

2025

Huazhong Cup Mathematical Modeling Challenge
Provincial Third Prize · Team Leader

2025

China Undergraduate Mathematical Contest in Modeling
Provincial Third Prize · Team Leader