It develops scalable techniques for detecting, explaining, and mitigating defects that are difficult to expose through conventional development practice alone. Such defects include concurrency errors, flaky and nondeterministic tests, and information-flow violations. This line of work draws on static analysis, systematic testing, automated repair, and empirical software engineering.
The goal is not only to find bugs after the fact, but also to improve software quality throughout the lifecycle. Prior work has shown that static-analysis-guided repair can use semantic summaries to fix concurrency defects while preserving program behavior. Other approaches combine testing and analysis to detect unstable behavior and flaky executions in event-driven systems, thereby improving the reliability of regression testing.
A central goal of this research is to support assurance in software systems whose scale, heterogeneity, and evolution make traditional methods insufficient. This includes event-driven, concurrent, and multilingual systems, where subtle interactions across components, languages, and execution contexts often undermine robustness. Scalable interprocedural analysis and summary-based reasoning help extend quality-assurance techniques to system scale. Empirical studies of nondeterministic failures further show the need for systematic exploration to expose latent defects.
By combining static reasoning with dynamic evidence from test execution, these techniques enable more effective diagnosis of unstable behavior, stronger regression guarantees, and automated or semi-automated fixes. Prior work on automated repair of security and correctness violations shows that program transformations validated by test suites can connect defect detection with remediation. More broadly, the aim is to make software quality assurance systematic, scalable, and better integrated with modern software development and maintenance.
Preferred student profile
Software engineering
Responsible researcher
Associate Professor Abhishek Tiwari
Associate Professor Adam Alami
Maximum number of students to host
Unlimited