KeynotesMSR 2022

Main Keynotes

Christian Kästner

From Models to Systems: Rethinking the Role of Software Engineering for Machine Learning

Building production systems with machine learning components is challenging and many projects fail when moving into production even when showing initial success with training machine-learned models. Unfortunately data science education focuses narrowly on data analysis, machine-learning algorithms, and model building but rarely engages with how the model may be used as part of a system. Engineering aspects beyond deploying models are often ignored or underappreciated, including requirements engineering, user experience design, planning and testing integration with non-ML components, and planning for evolution, leading to poor outcomes in many real-world projects. Software engineers and data scientists often clash in teams due to different goals, processes, and expectations, finding it hard to effectively coordinate and integrate work. In this talk, I argue for reshaping education and tools of data scientists and software engineers to shift from a model-centric to a system-centered view in their development activities. Attendees of MSR have a rare profile with expertise and a deep appreciation of both data science and software engineering. My hope is that we can use this expertise beyond data science in software engineering research, to better the education and tooling of both software engineers and data scientists. Let’s foster some engineering skills in data scientist students and let’s prepare our software engineering students with knowledge and tools for a world where they will almost inevitably interface with machine-learning components. Let’s prepare them for the interdisciplinary work needed to build production systems with machine learning components.

Bio: Christian Kästner is an associate professor and the director of the Software Engineering PhD program at the School of Computer Science at Carnegie Mellon University. His research originally focused on software analysis and the boundaries of modularity, especially in the context of highly-configurable systems. He also conducts research on sustainability of open source software and communities. His research often used data science methods and tools, such as when predicting how configuration options change the performance of a software system or when modeling the benefit of donations in open source projects, though he personally never cared much about data science as a topic in itself (and even though unlikely he wouldn’t mind another AI winter soon). In 2019 he started to co-teach a new course “Machine Learning in Production” at the intersection of software engineering and machine learning to better prepare the large number of students who, after graduation, start to work on software systems that integrate more and more machine learning (e.g., mobile apps, web applications, IoT devices). Since then, he also conducted research on collaboration, documentation, and quality assurance in teams where software engineers and data scientists interact.