About

Runqi is a postdoctoral researcher at the Oxford Internet Institute. Before joining the OII, he completed his PhD in Computer Science at the University of Sydney.

Runqi’s overarching pursuit is to develop trustworthy machine learning systems that deliver robust and human-aligned models. His previous research concentrated on understanding robust generalization to ensure reliable performance under worst-case scenarios, addressing key challenges such as learning flattened representations and mitigating undesirable overfitting. His current work focuses on human alignment in foundation models, aiming to ensure that they follow human instructions, conform to human values, and generate positive, responsible responses. He seeks to further advance artificial intelligence interpretability, ethics, and safety through a coordinated approach that combines red-teaming evaluations to uncover vulnerabilities with developer-teaming efforts to strengthen security mechanisms.

Runqi is always happy to connect and collaborate on topics related to trustworthy machine learning, and looks forward to advancing safe, human-aligned AI systems and promoting trustworthy AI for scientific discovery.

Research Interests

AI Safety; Trustworthy ML; Human-aligned Foundation Models; Trustworthy AI for Science

Positions at the OII

• Postdoctoral Researcher, March 2026 -