Towards Full-Lifecycle Quality Assurance of Modern Software
Speaker: 
Xiaoyu SUN
Lecturer and MComp Co-Convener
School of Computing
Australian National University
| Date: Time: Venue: | | 4 September 2026, Friday 2:00pm – 3:00pm School of Computing & Information Systems 2 (SCIS 2)
Level 2, Seminar Room 2-10
Singapore Management University
Singapore 178903
Please register by 3 September 2026 
|
|
About the Talk
Software has changed dramatically over the past decade. Modern software is no longer a self-contained program developed and operated by humans. It increasingly depends on complex software ecosystems, is deployed through highly automated cloud infrastructure, and is now being generated, operated, and even autonomously controlled by AI. These changes raise a fundamental question: how can we trust this new generation of software? In this talk, I will present our research towards full-lifecycle security assurance of modern software systems. The talk covers three stages of the software lifecycle: (1) understanding complex software behaviors through program analysis to uncover bugs and security risks before release; (2) automating the transition from software artifacts to reliable cloud deployments; and (3) governing agentic software at runtime to ensure that AI agents operate within intended security boundaries. Across these directions, I will discuss how program analysis, software testing, and LLM-based reasoning can complement each other to address challenges that are difficult to solve with any single technique. I will conclude with open questions and emerging research directions towards trustworthy AI-powered software systems.
About the Speaker
Xiaoyu SUN is a Lecturer in the School of Computing at the Australian National University. She received her Ph.D. from Monash University in 2023. Her research focuses on understanding, securing, and automating modern software systems across their full lifecycle, spanning mobile applications, cloud infrastructure, and AI agents. Her research has developed a series of techniques that extend the capabilities of program analysis to uncover software behaviors that were previously difficult to analyze. Several of these techniques have been integrated into FlowDroid, a widely used static analysis framework for Android. Building on this foundation, her recent work combines program analysis with LLM-based reasoning to automate software testing, cloud deployment, and the runtime governance of agentic software. Her work has been published at leading software engineering, and machine learning venues, including ICSE, FSE, ASE, TSE, TOSEM, and ICML.
