AI Red Teamer, LLM Generalist
Core
Stress-test large language models by designing creative, adversarial prompts to expose vulnerabilities in safety guardrails, bias, and robustness.
Role type
AI Red Teamer (LLM Generalist)
Builds
Adversarial prompts, harm taxonomies, and safety evaluations for frontier AI labs
Domain
AI Safety / LLM Security
Deliverable
production ML models
Required skills
Adversarial prompt engineering, jailbreak and evasion techniques, creative problem-solving, structured documentation, ethical judgment
Preferred skills
Python scripting, LLM API usage, structured data annotation, trust and safety experience, domain expertise in high-risk areas
Technologies
ChatGPT, Claude, Gemini, open-source LLMs, LLM APIs
Responsibilities
Craft multi-turn scenarios to stress-test AI guardrails across diverse risk categories; Discover ways around safety filters using jailbreak and prompt injection; Evaluate and score model responses against harm taxonomies; Document experiments and refine adversarial prompts; Collaborate with engineers and researchers to strengthen defenses
Seniority
Individual Contributor, hands-on experimentation