I am a research assistant in the Explainable AI group at the Fraunhofer Heinrich Hertz Institute in Berlin, supervised by Moritz Weckbecker, and a Master’s student in Computer Science at the Hasso Plattner Institute in Potsdam.
I work on backdoor detection for LoRA adapters using weight-space learning, and on understanding the backdoor mechanism itself with feature attribution methods. More broadly, I am interested in AI safety and interpretability, and in fine-tuning phenomena such as emergent misalignment and unexpected generalization.
Currently
- Research Assistant at Fraunhofer HHI, Explainable AI group (Dr. Sebastian Lapuschkin, Prof. Wojciech Samek) 2025 –
- M.Sc. Computer Science at Hasso Plattner Institute 2025 –
- Operations & Digital at START Berlin – building an internal platform for members and alumni. 2025 –
Previously
- Member of the appointment committee “AI Open Topic” at HPI. 2026
- B.Sc. Computer Science at HPI. Bachelor thesis on weight-space backdoor detection in LoRA adapters 2022 – 25
- Working student at SAP AI Core 2025
- Bachelor project with Amazon Berlin (AI and Sustainability chair, HPI) on LoRA merging techniques. 2024 – 25
- Data science project with Endress+Hauser 2024 – 25
- Member of the appointment committee “Data Systems” at HPI. 2024 – 25
- Working student at SPRYLAB – co-developed various applications, including Alpen Optics app 2022 – 24
Teaching
- Teaching Assistant for Algorithms & Data Structures at HPI. SS 2024
- Teaching Assistant for the introductory programming lecture at HPI. WS 2023/24