I am a PhD student and AI Security Researcher at MBZUAI in Abu Dhabi, UAE. My research focuses on semantic manipulation—examining how manipulating input data distributions can corrupt an agent’s reasoning without issuing malicious instructions. I am also interested in automated AI red-teaming methods, particularly in transforming existing capability benchmarks into adversarial environments to enable the systematic evaluation of agent resilience during capability assessments.
My background spans social sciences, psychology, machine learning, and cybersecurity. These multidisciplinary academic and professional experiences allow me, as a researcher, to approach problems from different perspectives and explore unconventional ways of thinking about AI security challenges.
Alongside my research, I actively participate in CTF competitions (@lmshunter), with a particular focus on AI and agentic systems. If you would like to learn more, you can visit my GitHub (@lmshunter) or explore the articles on my blog.
If you would like to discuss specific topics or collaborate on research, please feel free to contact me on LinkedIn.