I am a doctoral researcher at the Software Technology Group working on artificial intelligence for software engineering. My focus is on large language models (LLMs) for code generation, where I am trying to enhance the quality and security of the generated code through inference-time approaches such as retrieval or constrained decoding.
See also the interview about my work on the hessian.AI blog.
Bachelor/Master Theses: I am currently at full capacity and cannot supervise any additional theses. Please refrain from contacting me.
Short Biography
| when | what |
|---|---|
| since 2022 | PhD, Software Technology, TU Darmstadt |
| 2020 - 2022 | MSc, Computer Science, TU Darmstadt |
| 2016 - 2020 | BSc, Computer Science, TU Darmstadt |
Publications
2025
Benchmarking web API integration code generation.
Maninger, Chemnitz, Sharifloo, Brugger, Mezini,
AIWARE.
10.1109/aiware69974.2025.00034
Where do LLMs still struggle? An in-depth analysis of code generation benchmarks.
Sharifloo, Heydari, Kazerooni, Maninger, Mezini,
AIWARE.
10.1109/aiware69974.2025.00035
2024
Towards trustworthy AI software development assistance.
Maninger, Narasimhan, Mezini,
ICSE.
10.1145/3639476.3639770