About Me
I am a researcher at the SprintML Lab under the supervision of Franziska Boenisch & Adam Dziedzic at the CISPA Helmholtz Center for Information Security, and an M.Sc. student in Cybersecurity at Saarland University.
I generally work on topics related to Trustworthy Machine Learning, focusing on Large Language Models, Vision and Multimodality.
You can reach out to me via email at kaifs0351[at]gmail[dot]com.
Research Interests
Trustworthy Machine Learning; Privacy in ML; Natural Language Processing; AI Alignment; Human-Computer Interaction (HCI); Interpretability; Adversarial Machine Learning
Recent News & Updates
- [September 2026] Our paper "Benchmarking Membership Privacy Risks in Preference-Based LLM Post-Training" is accepted at the NeurIPS 2026 Evaluations and Datasets Track.
- [August 2026] Our paper "Design Theater: Evaluating the Gap Between User-Facing Design Reasoning and Implementation in Generative UI Tools" is accepted at the AAAI/ACM Conference on AI, Ethics, and Society (AIES) 2026.
- [April 2026] I started my M.Sc. in Cybersecurity at Saarland University, Germany.
- [March 2026] Our work "Auditing Preference-Based Post-Training of LLMs via Strong Membership Inference Attacks" is Accepted at the 3rd Workshop on Navigating and Addressing Data Problems for Foundation Models @ ICLR 2026
- [June 2025] Our work "Implementing Adaptations for Vision AutoRegressive Model" is Accepted at DIG-BUGS: Data in Generative Models Workshop @ ICML 2025
- [March 2025] Our paper on "Measuring Communal Violence Using Large Language Models" is accepted at the American Political Science Association Annual Meeting 2025. We will be presenting our work in Sept 2025.
Publications
- Kashif Imteyaz, Kaif Imteyaz, Nakul Rajpal, Kaif Shaikh, Michael Muller, Saiph Savage. Design Theater: Evaluating the Gap Between User-Facing Design Reasoning and Implementation in Generative UI Tools. In Proceedings of the AAAI/ACM Conference on AI, Ethics, and Society (AIES) 2026.
- Lorenzo Rossi, Kaif Shaikh, Franziska Boenisch, Adam Dziedzic. Auditing Preference-Based Post-Training of LLMs via Strong Membership Inference Attacks. In 3rd Workshop on Navigating and Addressing Data Problems for Foundation Models – ICLR 2026.
- Kaif Shaikh, Antoni Kowalczuk, Franziska Boenisch, Adam Dziedzic. Implementing Adaptations for Vision AutoRegressive Model. In Proceedings of the DIG-BUGS: Data in Generative Models Workshop — ICML 2025.