Define AI Security Methodology and Drive It Across the Practice: Define how AI security risk is assessed, monitored, and mitigated across product, engineering, and third-party AI integrations, with recognized depth on the current threat landscape: LLM workflows, agentic pipelines, MCP-based integrations, and attack classes including prompt injection, indirect injection, and tool-calling authorization gaps. You Will: Lead Threat Modeling and Product Security Reviews: Own threat modeling and product security review as the team''s primary upstream capability: build models from architecture and data-flow artifacts, derive concrete abuse cases and test scenarios, and drive security requirements into designs before they ship.