Red Teaming Domain Expert, Handshake

• This role requires creativity, curiosity, and an ability to think like an adversary while operating with strong ethical judgment
• No technical background is required
• What matters most is how you think, how you write, and how you problem-solve
• Strong hands-on experience using multiple LLMs
• Intuition for crafting prompts; familiarity with jailbreak or evasion techniques is a plus
• Creative, adversarial problem-solving skills
• Clear and thoughtful written communication
• Ability to tolerate emotionally heavy or graphic content
• Curiosity, persistence, and comfort with frequent failure in experimentation
• Strong ethical judgment and ability to separate adversarial thinking from personal values
• Self-directed, collaborative, and comfortable in feedback-heavy environments
• You go deep into unusual interests (fandoms, niche internet cultures, gaming exploits, Wikipedia rabbit holes, etc.)
• You come from a creative background, writers, visual artists, etc
• You are obsessed with AI and can’t stop talking about it
• Technical Requirements: Personal device running Windows 10 or macOS Big Sur 11.0+ and reliable smartphone access

Desirable

• This is a unique opportunity to join a fast-growing team shaping the future of AI through better data, better tools, and better systems—for experts, by experts
• As a Red Teamer, you will stress-test AI models by intentionally trying to break them
• Instead of checking whether an answer is correct, you’ll design creative, adversarial prompts that expose vulnerabilities—unsafe content, bias, broken guardrails, or unexpected behaviors
• Your work directly supports AI safety and model robustness for leading research labs
• This is a remote contract position with variable time commitments, typically 10–20 hours per week
• Crafting creative prompts and scenarios to intentionally stress-test AI guardrails
• Discovering ways around safety filters, restrictions, and defenses
• Exploring edge cases to provoke disallowed, harmful, or incorrect outputs
• Documenting experiments clearly, including what you tried and why
• Reviewing and refining adversarial prompts generated by Fellows
• Collaborating with engineers, tutors, and researchers to share findings and strengthen defenses
• Working with potentially disturbing content, including violence, explicit topics, and hate speech
• Staying current on jailbreaks, attack methods, and evolving model behaviors
• Schedule: Flexibility required, with some evening or weekend availability