I am a Doctoral Researcher at Saarland University and a member of ELLIS, working with Prof. Vera Demberg.
My research focuses on evaluating and improving the reliability of Large Language Models (LLMs). I am specifically interested in understanding the fundamental limitations of LLM architectures through expressivity frameworks and mechanistic interpretability and fixing them to improve LLMs’ reliability in practice.
Previously, I completed my Masters in Computational Linguistics at the University of Tübingen, where I worked with Prof. Detmar Meurers and Dr. Çağrı Çöltekin. Before that, I worked as a Research Assistant at IIT Madras with Prof. Mitesh Khapra.
Outside of the lab, I also help start-ups integrate GenAI into their products. In my free time, I enjoy hiking and reading poetry. I have been learning to play Violin for about a year now.
P.S. If you find me with my violin in my hand, you might want to look for some earplugs ;P
News
Nov 2025: Our paper on B-cos LM accepted at TMLR!
Oct 2025: Our paper showcasing recency bias of LLMs while reasoning accepted at IJCNLP-AACL 2025!
Sep 2025: Our work explaining effects pre-training and scale on architectural abilities of Transformers accepted at NeurIPS 2025!
Aug 2025: Our paper on improving factuality and attribution in multi-hop medical reasoning accepted at EMNLP 2025!
Sep 2023: Joined Saarland University as a PhD student in Prof. Vera Demberg’s group.