Artificial Intelligence Security Specialist USA

Citi•Jersey City, NJ

1d•Onsite

About The Position

Citi is seeking an Artificial Intelligence Security Specialist to join its Chief Information Security Office (CISO) teams. This role operates at the intersection of AI and cybersecurity, focusing on building defensive capabilities against AI-driven threats. The position involves solving novel problems in AI-driven vulnerability management, defining security architecture for AI deployment, securing AI agents that can mimic insider threats, and performing real security engineering. The work is not administrative but requires understanding code, architecture, and threats to find solutions. The role is based in the USA and offers opportunities to work on AI at enterprise scale, impacting over 180,000 employees and 30,000 developers.

Requirements

Hands-on LLM API experience (context management, tool use, evaluation, failure modes).
Agentic systems design.
AI safety at the infrastructure level.
Vulnerability research, exploit development, or pen testing with real depth.
Detection engineering for novel attack patterns.
Threat modeling (STRIDE, ATT&CK).
Security architecture.
Experience building and operating production systems.
Strong Python and/or systems programming.
Ability to digest dense technical research and turn it into actionable security recommendations.
Genuine curiosity.
Comfortable with ambiguity.
Biased toward building.
Able to work across disciplines.
Bachelor’s degree/University degree or equivalent experience.

Nice To Haves

Master’s degree preferred.
Comfortable reading disassembly or tracing through kernel code.
Published research, conference talks, or open-source contributions.

Responsibilities

Build defensive capabilities to counter AI-driven threats.
Develop pipelines for AI-driven vulnerability management, focusing on triage, verification, and remediation.
Define security architecture for the deployment, monitoring, and governance of agentic AI in a global bank.
Develop standards, evaluation frameworks, and production runtime monitoring for AI.
Secure AI agents that can behave like insider threats, including harvesting credentials, escaping sandboxes, and adapting to monitoring.
Design containment architectures and runtime controls for safe enterprise-scale operation of AI agents.
Understand code, architecture, and threats to engineer solutions.
Engage in AI-assisted penetration testing and automated exploit validation.
Bridge the gap between AI vulnerability discovery and application team remediation.
Define how the bank deploys AI safely, providing security architecture and assurance for new implementations.
Build next-generation AI-powered tools for CISO colleagues.
Test new models at the cutting edge of creation and influence.
Own AI products CISO depends on in production, including security assurance, cyber security operations, governance, and controls.
Ensure the reliability of AI products and evolve them rapidly.
Develop detection, triage, and response strategies for a world where adversaries use AI.
Implement behavioral analytics for AI agents.
Create playbooks for AI-originated attack scenarios.