Automated testing of computer games presents unique challenges due to their complexity, dynamic nature, and pervasive randomness, making traditional static testing approaches fragile and inadequate for thorough exploration or bug detection. To provide context, this talk will begin with an overview of the current landscape of automated test generation techniques used for games. Within this landscape, modern approaches like Reinforcement Learning have shown promise in game playing; however, these agents are usually trained on game-specific objectives (winning) over systematic bug exploration, demanding extensive training and hand-crafted reward functions that struggle with frequent game changes during development. For this reason, we argue against abandoning traditional automated test generation in the context of game testing. Instead, we propose a hybrid approach that combines white-box coverage-guided test generation with reinforcement learning in terms of neuroevolution. This methodology generates dynamic test suites—–machine learning models that learn to play games implicitly, driven by the goal to explore their source code and to maximize code coverage.

Gordon Fraser is a full professor in Computer Science at the University of Passau, Germany. He received a PhD in computer science from Graz University of Technology, Austria, in 2007, worked as a post-doc at Saarland University, and was a Senior Lecturer at the University of Sheffield, UK. The central theme of his research is improving software quality, and his recent research concerns the prevention, detection, and removal of defects in software. He is known for his work on search-based testing, in which he created popular unit test generators such as EvoSuite or Pynguin.

Over the past decade, my research has been dedicated to the analysis of mobile applications, with a primary focus on the Android platform. As we delve into the intricacies of mobile app quality and trustworthiness, it becomes imperative to question the prevailing bias in our research toward the Android platform. This talk aims to shed light on the reasons behind this predominant focus and provide insights into the shift of my research paradigm towards the iOS platform. Specifically, the talk will illuminate some of the unique challenges associated with analyzing iOS applications, emphasizing the differences in terms of collecting apps for analysis, the security features, and app deployment mechanisms which make iOS and Android different. I will showcase key findings of the prototype that we developed to identify third-party libraries in iOS apps, offering a comparative lens to understand the key differences between these platforms.

Alessandra Gorla is an Associate Research Professor (tenured) at the IMDEA Software Institute in Madrid, Spain. She received her Bachelor’s and Master’s degrees in Computer Science from the University of Milano-Bicocca, Italy, and her Ph.D. in Informatics from the Università della Svizzera Italiana (USI), Switzerland, in 2011. Her Ph.D. introduced the concept of Automatic Workarounds, a self-healing technique for recovering web applications from field failures, for which she received the Fritz Kutter Award for the best industry-related Ph.D. thesis in Computer Science in Switzerland. Before joining IMDEA in 2014, she was a postdoctoral researcher at Saarland University, Germany, and a visiting researcher at Google in Mountain View, California. Her research focuses on software testing, software analysis, software security, and AI for software engineering, with a particular emphasis on developing automated techniques to improve the reliability, security, and trustworthiness of modern software systems. She is the recipient of the Ramón y Cajal Fellowship, the Spanish Data Protection Agency’s Emilio Aced Award, a Facebook Research Award, and the 2025 ASE Most Influential Paper Award.

The fundamentals of software and software engineering are undergoing a significant transformation. This talk introduces AIware (AI-Powered Software) and explores the emergence of Agentic Software Engineering. We will discuss how to move beyond informal “vibe coding” toward “vibe engineering,” and ultimately to “Agentic SE,” a more disciplined and powerful framework for creating production-grade software. The talk explores a future where the developer’s role evolves from simply writing code to becoming an orchestrator, collaborator, and mentor for AI teammates, as we move beyond the era of simple “copilots” into a more dynamic, collaborative partnership between humans and AI. We ground this vision in findings from our analysis of AIDev, a large-scale dataset of roughly 10 million agent-generated pull requests. The analysis highlights what real agentic work looks like in the wild, revealing both the potential and the practical challenges that lie ahead. Join me to discover how we can build systems of greater complexity and scale through conversation, intent, and creative partnership with AI. This is a chance to understand the foundational principles of this emerging field and to prepare for the next revolution in software engineering.

Ahmed E. Hassan is a Mustafa Prize Laureate (a Nobel-level recognition) and a Fellow of ACM and IEEE. For over 25 years, he has been a driving force behind the integration of AI into Software Engineering research and practice worldwide. His open-access book, Agentic Software Engineering, defined this emerging era and continues to shape it.

Picture the high-stress software engineering scenario when malware is developed to exploit a flaw in your software project. The malware is spreading, causing damage or leaking information from your clients. A team of human analysts must find the malware, understand how it works, and deploy mitigating procedures. This team will rely on a myriad of AI-enabled agents such as network monitors and source and bytecode analyzers. They will design fixes, often with the help of AI code generation tools, etc. Today, this process can be largely ad hoc, with different tools deploying different technologies at different times. Yet, there is a vision of coordinated, human-AI collaboration in which AI agents anticipate human information needs, and people communicate with these agents through natural language and other human-centered means. In this talk, I argue that insights from computational neuroscience, particularly models of human attention, cognitive load, and error patterns, can enable AI agents to better anticipate when and how humans will intervene. By modeling where developers look, what they miss, and how they recover from errors, we can design agents that adapt to human cognitive states, proactively surface risks, and collaborate more effectively. I present a future in which software engineering AI systems understand not just code, but the humans who write and reason about it.

Collin McMillan is a Professor in the Department of Computer Science and Engineering at the University of Notre Dame. His professional research area is Software Engineering, with a special emphasis on Artificial Intelligence for Software Engineering (AI4SE). He studies the “concept assignment problem in program comprehension”, which is the mismatch between high-level concepts that programmers intend for software behavior, and the low-level implementation details embodied by computer source code and machine systems. He is known for his research strategy in which his group: 1) studies what people do during program comprehension, 2) designs models to mimic what they do, and 3) applies these models to improve AI-based program comprehension tool support. Dr. McMillan’s work has been recognized with the NSF CAREER award, the ASEE Illinois/Indiana Teacher of the Year award, and several best paper and distinguished paper awards.

Autonomous driving systems (ADS) require extensive testing to ensure safety, reliability, and robustness across a wide spectrum of operational conditions. This talk introduces the state of the art in testing and evaluation of ADS, with a particular focus on simulation-based techniques and the challenges of transferring results from simulation to real-world environments. We discuss approaches for scenario generation and performance assessment, as well as current limitations in realism, sensor modeling, and non-deterministic behavior. Finally, the talk explores emerging research directions, such as sim2real consistency analysis, uncertainty quantification, and AI-driven test generation.

Andrea Stocco is an Assistant Professor at the Technical University of Munich at the Chair of Software Engineering for Data-intensive Applications of the School of Computation, Information and Technology. He is also the head of the Automated Software Testing unit at fortiss. His research focuses on the interface between software engineering and deep learning with the goals of improving the robustness, reliability, and dependability of data-intensive software systems. He is the recipient of several awards, including a Distinguished Paper Award at the 33rd IEEE International Conference on Software Analysis, Evolution and Reengineering (SANER 2026), and two Distinguished Paper Awards at the 18th International Conference on Software Testing, Verification and Validation (ICST 2025). He serves on the program committees of top-tier software engineering conferences such as ICSE, FSE, ASE, ISSTA and ICST, and reviews for numerous software engineering journals including TSE, TOSEM, EMSE, JSS, and IST.

In recent years, Mining Software Repositories (MSR) research has produced a large body of empirical evidence on the relationships between software development practices, code characteristics, and software quality outcomes. However, most existing studies remain fundamentally correlational, making it difficult to reason about causality and limiting the reliability of the conclusions drawn from observational data. While controlled experiments represent the gold standard for causal inference, they are often infeasible in real-world software engineering contexts due to cost, scale, ethical, or practical constraints. This talk explores the use of cohort studies, a methodology widely adopted in epidemiology and other observational sciences, as a practical alternative for strengthening causal reasoning in MSR research. It introduces the principles behind cohort studies, discusses the methodological challenges of applying them to software repositories, and presents a set of heuristics for designing robust cohort-based mining studies. To illustrate the approach, the talk includes a hands-on exercise in which participants design their own cohort study for an MSR research question.

Sira Vegas is Full Professor of Software Engineering at the Universidad Politécnica de Madrid, Spain. Her research focuses on experimental software engineering, AI for software engineering (AI4SE), and software testing. She has published over 100 articles in international journals and conferences, participated in national and international research projects, and collaborated with leading researchers worldwide. She serves regularly on program committees and as a reviewer for top venues such as IEEE TSE, ACM TOSEM, EMSE, ICSE, ESEM, MSR and ASE. Sira has co-chaired several international conferences (ESEM, EASE, PROFES, ICSE-DS, ICPC-RENE, SANER-JF) and currently chairs the ISERN Steering Committee. She holds a Ph.D. from UPM and has been a visiting researcher at the University of Maryland, Fraunhofer IESE, University of Oulu, and Tongji University.