I have very broad interests in Software Engineering and Artificial Intelligence but in particular focus on Empirical Software Engineering, Mining Software Repositories, AI4SE and SE4AI.
Where the research programme is heading over the next funding cycle.
Moving from LLM-assisted code recommendation to autonomous coding and security agents whose outputs can be verified and repaired automatically — the direction underpinning the VulGuard-Agent proposal. Targets ARC Discovery/Linkage and industry co-funding (e.g. cloud/security partners).
Extending feature-based and LLM-based pattern detection/summarisation into full architecture-recovery tooling, so teams can inventory reuse and technical debt at scale. Builds on the CodeLabeller/CFS/PRST line of work, targeting Discovery Early Career and industry-linkage schemes.
Rigorous, replication-ready studies of how AI coding assistants change developer behaviour, code quality and maintainability — mixed-methods work suited to ARC Discovery and international collaboration grants.
Translating AI4SE and MSR findings into curriculum and assessment design, connecting the research programme to the teaching and curriculum-review work already underway. A natural fit for OLT/teaching-innovation grants alongside the research-council schemes above.
Tip: Selecting an area filters the projects below.
Click All areas to reset the list.
Advancing evidence‑based software practice through rigorous empirical methods—large‑scale mining, experiments and mixed‑method studies of developer workflows, artefacts and socio‑technical systems—to evaluate and improve tools and processes.
Analysing version control, issue and code review repositories, Q&A archives, CI/build servers and runtime telemetry with data science, ML/AI and qualitative methods to surface actionable insights that improve software engineering practice and guide project evolution.
Advancing AI‑driven Software Engineering through human‑centred, trustworthy, sustainable and collaborative methods - spanning LLM‑enabled automation, AI‑assisted design, recommender systems for code and repair, prompt engineering, and rigorous efficacy measurement beyond traditional metrics.
Engineering reliable, reproducible AI‑enabled systems—spanning SE for models and AI‑infused systems; AI code, libraries and datasets; autonomic and self‑healing systems with automated model repair; rigorous testing, verification, validation and user‑based evaluation; and requirements engineering.
A feature-based and LLM-based design pattern summarisation system that parses Java systems with JavaParser, produces JSON knowledge graphs and turns them into readable English generated narratives.
A machine-learning model that detects Gang-of-Four patterns inside Java projects, combining structural fingerprints with semantic cues so teams can inventory reuse opportunities.
A web-based annotation environment where researchers and practitioners label Java design pattern instances and summaries to bootstrap supervised learning datasets.
The first effort to blend supervised learning with crowdsourcing for summarising source code fragments harvested from Eclipse and NetBeans FAQs.
A PageRank-inspired approach for summarising duplicate-heavy bug reports drawn from Eclipse, Mozilla, KDE, and Gnome tracker conversations.