I am a PhD student at Aarhus University in Denmark. My main field of research is runtime verification (RV) applied to self-adaptive systems, or more generally, RV applied to systems that undergo changes after deployment. RV is about applying formal verification to system behavior at runtime. In layman’s terms, we can use RV to give stronger guarantees about system behavior because the methods are mathematically based and can reason about it more rigorously. Traditionally, the RV specifications used to verify system behavior are fixed, and it is often assumed that the system requirements are stable. I challenge that assumption with my research.

During my PhD, in close collaboration with my colleague Thomas Wright, I have developed the first stream-based RV language capable of reconfiguration at runtime, where historical states are preserved when sensible. The hypothesis is that our language is better suited for dealing with deployed systems where requirements may change. During my research, I hope to demonstrate the efficacy of our language in collaboration with Andrew Putman and Peter Lewis. Here, we will carry out a transfer-learning experiment on healthcare data, iteratively fine-tuning the model as we explore several new fairness metrics. We use RV to guarantee that the model does not regress on the fairness metrics it was previously fine-tuned to satisfy.