AI Red-Teamer — Adversarial AI Testing

About the Role

We are building an advanced AI red team to proactively test and stress-test conversational AI systems. This role focuses on adversarial evaluation—probing models with carefully designed inputs to uncover vulnerabilities before they reach production.

All work is text-based and may involve sensitive topics such as bias, misinformation, or harmful behaviors. Participation in higher-sensitivity testing is optional and supported with clear guidelines and wellness resources. Topics are always disclosed in advance.

Responsibilities

• Red-team conversational AI systems through:
• Jailbreaks and prompt injections
• Misuse and abuse scenarios
• Bias exploitation
• Multi-turn manipulation and agent testing
• Generate high-quality human data:
• Annotate model failures
• Classify vulnerabilities
• Flag systemic and recurring risks
• Apply structured methodologies:
• Use taxonomies,...