Lead AI Security Engineer — Agentic Platform Defense

EY; Dallas (TX); Remote; USD 140.000 - 190.000; Vollzeit

Zusammenfassung

EY is seeking an AI Security Engineer to own the security posture of our Agentic AI platform end to end. You will tackle threat modeling, secure-by-default governance, and sandboxed execution across cloud-native stacks.

The role demands depth in AI security, red-teaming, policy-as-code, and collaboration with architecture to define agent authority and supply-chain integrity for regulated environments.

Qualifikationen

• Experience with cloud-native security and AI/agentic threat models.
• Ability to write threat models and defend engineering to CISOs or regulators.
• Strong background in red-teaming and adversarial testing.

Aufgaben

• Own the platform threat model across deployment targets from cloud to air-gapped.
• Define security engineering and control sets for all platform layers including Kubernetes and secrets.
• Set secure-by-default contracts for agent templates and network policy.
• Lead defense against agentic threat classes including prompt injection and hijacking.
• Define the agent authority model with delegation and consistency constraints.
• Own the sandboxing security standard for agent-generated code execution.
• Secure the model and knowledge supply chain, including provenance and integrity.
• Secure agent-to-agent and tool protocols and inter-agent communications.
• Lead AI red teaming and adversarial testing.

Kenntnisse

Threat modeling

Cloud-native security

Red teaming

Policy-as-code

Threat hunting

Tools

Kubernetes

Docker

Cloud platforms

Jobbeschreibung

EY is seeking an AI Security Engineer to own the security posture of our Agentic AI platform end to end. You will tackle threat modeling, secure-by-default governance, and sandboxed execution across cloud-native stacks.

The role demands depth in AI security, red-teaming, policy-as-code, and collaboration with architecture to define agent authority and supply-chain integrity for regulated environments.