About me

I am a fourth-year PhD student in the Multilinguality and Language Technology lab at the German Research Center for Artificial Intelligence (DFKI), supervised by Josef van Genabith and Simon Ostermann.

In my PhD thesis, I study what happens inside large language models when they solve challenging tasks. LLMs often look impressive, but their performance frequently rests on simple heuristics that make them brittle. Using interpretability methods, I trace the mechanisms behind these shortcuts to understand why they form and how to overcome them.

Beyond my thesis, I am broadly interested in understanding and controlling what happens inside language models:

  • Mechanistic interpretability: reverse-engineering the internal mechanisms behind model behavior, and using these insights to improve models
  • Tokenization: how the choice of tokens shapes what a model has to learn and how it represents it
  • Reasoning mechanisms: how LLMs solve arithmetic and other reasoning tasks internally

News

Master Theses

I am always looking for talented Master students of Language Science and Technology or Computer Science, who want to work on a thesis in the area of interpretable, robust, or trustworty NLP.

If you are interested, feel free to reach out with your own research question related to my research interests above.

When contacting me, please include:

  • Your CV
  • Your Transcript of Records
  • A short research question pitch (possibly accompanied with relevant literature that your idea builds on)