AI Safety Red Teamer

This is a fully remote position, open to applicants in United States.

📋 Description

• Create adversarial prompts to thoroughly evaluate frontier AI models.

• Detect jailbreaks, unsafe behaviors, hallucinations, and policy violations.

• Assess model resilience across domains such as misinformation, cyber threats, biosecurity, fraud, political content, and other sensitive areas.

• Record vulnerabilities and assist in the preparation of safety benchmarking and red-teaming reports.

• Work collaboratively with AI researchers to enhance model alignment, robustness, and safety.

• Undertake projects as an independent contractor based on project requirements and performance.

⛳️ Requirements

• A Bachelor’s degree or higher in Computer Science, Cybersecurity, Journalism, Communications, Psychology, Biology, Chemistry, Public Policy, or a similar field.

• Over 5 years of professional experience in AI Safety, AI Red Teaming, Trust & Safety, cybersecurity, investigative journalism, life sciences, or a related area.

• Exceptional analytical reasoning, prompt design, and written communication abilities.

• Proven experience in designing adversarial prompts or assessing frontier AI systems.

• Preferred: experience in AI Red Teaming, RLHF, SFT, AI Alignment, or Trust & Safety.

• Preferred: knowledge of jailbreak testing, prompt engineering, or adversarial evaluation techniques.

• Preferred: specialization in grey-area fields including cyber threats, biosecurity, political content, misinformation, or scientific safety.

• Ability to work as an independent contractor is essential.

• H1-B and STEM OPT candidates are not eligible for support.

🏝️ Benefits

• Fully remote position.

• Flexible working hours to accommodate your schedule.

• Weekly payments through Stripe or Wise.

• Competitive salary.

• Opportunity to collaborate with top AI researchers and safety teams.

• Chance to impact the safety and alignment of frontier AI technology.

• Referral bonus of up to $340 for each successful referral.