Frontier AI Safety Red Teamer (Remote)

Mercor; Dublin; On-site; EUR 83,000 - 100,000; Full time

Job summary

Mercor is seeking an AI Safety Red Teamer for a contract engagement. You design adversarial prompts to stress-test frontier AI models, identify jailbreaks, unsafe behaviors, and policy failures, and document vulnerabilities for safety benchmarking.

In collaboration with AI researchers, you will help improve model alignment, robustness, and safety through rigorous evaluation across sensitive domains such as misinformation, cyber, biosecurity, and political content.

Qualifications

• Bachelor-level or higher in a related field and 5+ years in AI safety or red teaming.
• Proven ability to design and analyze adversarial prompts for frontier AI systems.
• Strong written and verbal communication skills.

Responsibilities

• Design adversarial prompts to stress-test frontier AI models.
• Identify jailbreaks, unsafe behaviors, hallucinations, and policy failures.
• Evaluate model robustness across misinformation, cyber, biosecurity, fraud, political content, and other domains.
• Document vulnerabilities and contribute to safety benchmarking and red-teaming reports; collaborate with researchers.

Skills

Analytical reasoning

Prompt design

Written communication

Adversarial prompt design

Education

Bachelor's degree in Computer Science

Bachelor's degree in Cybersecurity

Bachelor's degree in Journalism

Bachelor's degree in Communications

Bachelor's degree in Psychology

Bachelor's degree in Biology

Bachelor's degree in Chemistry

Bachelor's degree in Public Policy

Job description

Mercor is seeking an AI Safety Red Teamer for a contract engagement. You design adversarial prompts to stress-test frontier AI models, identify jailbreaks, unsafe behaviors, and policy failures, and document vulnerabilities for safety benchmarking.

In collaboration with AI researchers, you will help improve model alignment, robustness, and safety through rigorous evaluation across sensitive domains such as misinformation, cyber, biosecurity, and political content.