Amir Nuriyev

Focus on AI security and side-channel research: Amir's work centers on understanding how non-text signals—such as expert routing decisions and physical chip signatures—can leak sensitive information from modern model architectures, and on developing measurements and mitigations for these leakage modes. matsprogram+2

Active contributor in recent AI-safety research fora: Amir has presented and discussed this work at community and workshop venues (e.g., MATS symposium and the AVID Workshop), indicating engagement with the AI-assurance and verification community. substack+2

Early-career researcher with both academic and practical experience: While an undergraduate at MBZUAI, Amir has combined coursework with internships and teaching roles (e.g., Kazakhtelecom data-science internship and AddisCoder TA), reflecting a blend of research, applied, and educational activity. digitaloceanspaces+1

Collaborative, program-based mentorship and training: Participation in structured programs (MATS cohorts and extension, SPAR) and listed advisor relationships indicate Amir develops work within mentor-guided research programmes focused on alignment, governance, and security. matsprogram+2