Red Team Engineer

• Design and execute red teaming attacks on frontier AI systems, identifying vulnerabilities and translating findings into clear reports.
• Design new challenges and adversarial scenarios for ongoing Arena competitions.
• Conduct red teaming engagements with external clients covering prompt injection, tool use, and agent exploitation.
• Build internal tools and processes for the Arena including cheating detection and break verification.

This role directly contributes to the safety and security of AI systems by identifying vulnerabilities before they can be exploited, helping to build more robust and trustworthy AI technologies.

• Experience in red teaming, security research, or adversarial machine learning.
• Strong understanding of AI systems, including LLMs and agentic architectures.
• Proficiency in Python and familiarity with AI/ML frameworks.
• Excellent communication skills for translating technical findings into clear reports.

Competitive compensation and benefits package. Opportunity to work at the forefront of AI safety with a mission-driven team.

Gray Swan is an AI safety company focused on red teaming and security research to ensure the safe deployment of frontier AI systems.