(2/2) Sakina devised novel ways, relying in part on Large Language Models and other ML techniques, to predict, classify (according to root causes), and automatically fix flaky test cases. Since flaky tests are such an acute problem in many industrial contexts, this topic is particularly important

(2/2) His work covers: (1) At design time, identify combinations of learned component outputs and operational contexts that could lead to (system-level) hazards, (2) At runtime, predict learned component outputs that can lead to system hazard, given its operational context.

(2/2) This is particularly important in the context of continuous integration. She relied on evolutionary computing and Large Language Models, among other things.