Automated program repair aims to automatically change the source code of a given program to fix a given bug. This area made significant advances in recent years, and companies such as Meta and Bloomberg have succesfully deployed program repair systems to support their development workflows. However, current program repair techniques still do not scale to large code bases, often output wrong patches, and produce results that are hard to interpret, which limits their wider adoption.
In this project, you will work towards addressing these limitations. This will be done by making program repair property-driven, where the property is defined via program analysis such as Datalog-based static analysis and abstract interpretation. You will also investigate in what ways developers can benefit most from automatically generated patches in realistic industrial settings.
Applications from the UK are invited for a funded 4-year PhD studentship at UCL Computer Science. The starting date is either January or September, depending on when a suitable candidate is found. The project will be done under the supervision of Dr Sergey Mechtaev. Applicants should:
The scholarship covers the Home UCL registration fees, the standard UKRI level of stipends and a budget over 4 years to cover the student’s research expenses (eg. laptop, conference attendance costs, etc.).
Interested applicants should send their CVs and trascripts to Dr. Sergey Mechtaev at firstname.lastname@example.org