Computational Intelligence for Software Engineering Lab

Our research focuses on the exciting intersection between software engineering and the emerging AI technologies. We study state-of-the-art AI techniques for software engineering problems, and have world-leading expertise in automated debugging, testing, and agentic solutions. We also bring our software engineering expertise to the new generation of AI-based software systems, and lead the work on designing and testing of LLM-based agents.

Recent Updates

Atropos is accepted to ISSTA 2026

Atropos can perform early terminations of unsuccessful LLM inferences; further, it can migrate existing contexts to stronger model to save the inference. [more...]

Clotho is accepted to FSE 2026

We can now measure task-specific surprise adequacy for LLM inputs. [more...]

AutoCrashFL is accepted to ICSE SEIP 2026

The latest collaboration on large scale crash localisation between COINSE and SAP has been accepted into ICSE SEIP 2026. [more...]

AutoSD accepted at EMSE

Automated Scientific Debugging research accepted into EMSE. [more...]

ICSME 2024 Best Indistry Paper Award

COINSE/SAP collaboration wins Best Industry Paper award at ICSME 2024. [more...]

Flaky Symptoms at ICSME 2024

Our paper about detecting flaky test executions will appear at ICSME 2024. [more...]