Track 2: Predictive Software Testing
This track focuses on creating an intelligent and adaptive testing ecosystem capable of safeguarding quality in massive, rapidly-evolving codebases. As software scales, traditional testing methods often break down due to long regression cycles and high infrastructure costs. We will explore (1) learning-based techniques (e.g., LLMs, agents) to capture the intent behind code changes, (2) defining new metrics for test “quality” that go beyond code coverage, (3) predicting the success of code changes based on test quality, providing developers quick feedback, and (4) designing autonomous agents that automatically refactor, prune, and optimize test code, preventing test decay.
Track Leaders TU Delft
Track Lead
Track Lead
Track Leaders Meta
Track Lead
