Sergey Vlasov is a Senior Software Engineer at Radware, where he works at the intersection of network security and AI infrastructure. With nearly 12 years of security engineering experience, he focuses on architectural patterns that fail predictably — and how to eliminate those failure modes by construction rather than detection. His independent research project, Agent Hypervisor, proposes execution governance as a missing layer in agentic AI security: defining what actions exist in an agent's world rather than monitoring what actions it takes. This work does not represent Radware's position.