Staff Product Security Engineer

Okta is The World's Identity Company, focused on securing identities across various technologies. The Staff Product Security Engineer will be responsible for identifying and resolving security risks related to AI systems, conducting research, and mentoring other engineers in the field of AI security.

Responsibilities:

• Conduct offensive security research focused on agentic AI systems: prompt injection, agent privilege escalation, tool-binding abuse, and agentic supply chain attacks against internal developer platforms
• Perform security assessments of Okta's AI platforms—including agentic systems and LLM-integrated products—across design, code, and runtime
• Build reusable security tooling that multiplies the entire Product Security team's capability
• Run the AI security vendor and tooling evaluation program: design and operate a benchmarking harness against AI security tools
• Perform manual code review of AI and agent-based system implementations across multiple languages
• Develop threat models for agentic architectures, orchestration layers, and LLM-integrated services
• Translate research findings into actionable guidance for engineering teams building AI-powered features and platforms
• Represent Okta externally through security research, conference presentations, white papers, and publications
• Mentor engineers across Product Security on AI/agentic security concepts, tooling, and assessment methodology

Requirements:

• 7+ years of experience in information security, with meaningful depth in application security, offensive research, or AI/ML security
• Demonstrated hands-on experience assessing LLM-integrated systems and agentic AI architectures—not just familiarity with the concepts, but evidence of having found real vulnerabilities in them
• Strong offensive mindset: the ability to model what an adversary does with an agentic system, identify where the model's reasoning or the orchestration layer breaks down, and construct scenarios that make the risk concrete
• Experience building security tooling and automation—scripts, scanners, detection logic, or evaluation harnesses—that other engineers actually use
• Proficiency in at least two programming languages (Python and one of: Go, Java, TypeScript, C/C++)
• Advanced experience in threat modeling, manual code review, and penetration testing, applied to complex distributed systems
• Knowledge of authentication and authorization protocols (OIDC, OAuth 2.0, SAML) and their implementation risks
• Strong communication skills: the ability to write clearly for technical and non-technical audiences, document research findings with precision, and present at external venues
• Experience producing external security research—publications, conference talks, blog posts, or open-source tooling
• Familiarity with agentic framework internals (tool-use protocols, MCP, function-calling patterns, agent orchestration architectures)
• Experience with SAST, DAST, SCA, and fuzzing tooling applied to AI/ML pipelines or CI/CD systems
• Strong cryptographic knowledge and experience in identifying cryptographic implementation flaws
• Ability to develop proof-of-concept exploits that demonstrate AI/agentic vulnerabilities to engineering and product leadership
• Experience contributing to security standards, SDL processes, or vulnerability research programs