What happens when code changes?

My research focuses on specific tasks within code review: understanding what a code change does, which parts of a system it affects, and how tools such as change impact analysis and change classification can support developers in these tasks.

Research interests

01

Code review

The setting my work applies to: supporting specific review tasks, such as understanding a change and spotting the follow-up changes it needs.

02

Change impact analysis

Finding which code a change affects, and describing what kind of change it is.

03

LLMs for software engineering

Testing what large language models can really do for code analysis and build failures, with honest datasets and evaluations.

04

Empirical software engineering

Studying software and developers through data, experiments and user studies.

Publications

Google Scholar ↗
  1. AgenticDev 2026 1st International Workshop on Agentic AI for Next-Generation Software Development, ASE 2026

    katIA: Hybrid Source Code Change Impact Analysis (Extended Abstract)

    Katharina Stengg, Christian Macho, Martin Pinzger

  2. Under revision Journal of Systems and Software (JSS)

    Can Open-Weight LLMs Replace Build Log Analyzers?

    Christian Macho, Katharina Stengg

  3. SANER 2026 IEEE International Conference on Software Analysis, Evolution and Reengineering, ERA track

    Agent-Based Dependency-Related Build Repair

    Christian Macho, Katharina Stengg, Martin Pinzger

  4. arXiv 2025 Preprint

    A Dataset and Preliminary Study of Using GPT-5 for Code-change Impact Analysis

    Katharina Stengg, Christian Macho, Martin Pinzger

    Can GPT-5 tell which code entities a change impacts? We built a dataset of seed changes and impacted entities and found that today’s models still struggle, and that giving them the diff helps a little.

  5. Master’s thesis 2025 University of Klagenfurt

    Predicting Build Results using Software Metrics and Developer Sentiment

    Katharina Stengg

Talks, posters & events

  1. Poster ASE 2026 · AgenticDev workshop

    katIA: Hybrid Source Code Change Impact Analysis

  2. Guest lecture Winter semester 2026/2027 · University of Klagenfurt

    Code Review in the Era of AI

    In Christian Macho’s course Advanced Software Engineering

  3. Summer school August 2025 · USI Lugano, Switzerland

    SIESTA 2025, International Software Engineering Summer School

Let’s collaborate

I’m always interested in collaborations, especially where software engineering meets other fields: code review, human factors in software development, and biomedical engineering, for example using biometric signals to understand how developers read and review code.

Have an idea or want to work together? Email me at katharina.stengg@aau.at.