I am a PhD Candidate at CISPA Helmholtz Center for Information Security, working in the SprintML lab under the supervision of Dr. Adam Dziedzic and Dr. Franziska Boenisch.
My research focuses on the privacy and security of large foundation models (LLMs). I am currently focusing on privacy of multimodal models (paper coming soon!).
Previously, I studied at Wrocław University of Science and Technology (MEng in Trusted AI, BEng in Cybersecurity, both with honors), and worked on multilingual audio DeepFake detection and biometric systems.
- Privacy in Machine Learning
- LLM & Multimodal Model Security
- Audio DeepFake Detection
Education
- Final grade: 5.5 (top score).
- Graduated with honors.
- Thesis: Multilingual aspects of Audio DeepFakes.
- Rector’s scholarship holder during the whole studies.
- Final grade: 5.5 (top score).
- Graduated with honors.
- Thesis: Verification and identification behavioural biometric system based on typing errors
- Rector’s scholarship holder during the whole studies.
Experience
Member of the SprintML lab — Secure, Private, Robust, INterpretable, and Trustworthy ML. Supervised by Dr. Adam Dziedzic and Dr. Franziska Boenisch. Research focus:
- Privacy and security of large foundation models.
Member of the SprintML lab. Supervised by Dr. Adam Dziedzic and Dr. Franziska Boenisch.
- Conducted research on empirical privacy protection for LLM adaptations.
- Automation of security processes.
- Co-creation of the vulnerability management process for the hybrid cloud infrastructure using Orca Security.
- Creating dashboards using MS PowerBI and Grafana.
Technologies: Python, Orca Security, Nessus, PowerBI, Grafana, SQL
- Course delivery: Introduction to Python for the security domain.
- Risk assessment and threat modeling.
Technologies: MITRE ATT&CK, Risk Assessment