Hi, I'm Roman, a researcher in AI Safety at The AI Lab at The American College of Greece. This academic year, I'm graduating with a BSc (Hons) in Cybersecurity. After graduation, I plan to continue my academic and research career as a Ph.D. student.

My research focuses on AI alignment, trustworthiness, and security, to minimize the risks associated with the inevitable deployment of AI systems in critical domains including healthcare, cybersecurity, and robotics.

Currently, under the supervision of Prof. Ioannis T. Christou, I'm developing computer vision models for early diagnosis of acute lymphoblastic leukemia, and verifying the accuracy of learned representations based on model activations with gradient-weighted class activation mapping (Grad-CAM). Moreover, I'm actively developing my thesis on alignment assessment of vision-language-action (VLA) models with respect to safety regulations.