AI Safety Red Team Specialist (Remote)

Obsidian; New York (NY); Sur place; USD 90 000 - 130 000; Plein temps

Avantages offerts par ce poste

Experience in human data-driven AI red teaming

Role in enhancing AI safety and trustworthiness

Résumé du poste

Obsidian is assembling a red team for AI model security, specifically focusing on identifying vulnerabilities and generating quality datasets. Candidates should have prior red teaming experience and a curiosity for probing AI systems. The role involves working with adversarial techniques to ensure AI safety.

Success in this position means highlighting risks AI models may overlook, enabling clients to strengthen their systems. This role offers an impactful position in a cutting-edge field, contributing to safer AI applications.

Qualifications

• Experience in AI adversarial work, cybersecurity, or socio-technical probing.
• Ability to push systems to their limits.
• Familiarity with frameworks or benchmarks for testing.

Responsabilités

• Red team conversational AI models and agents, identifying biases and misuse cases.
• Generate and annotate high-quality human data, documenting findings.
• Follow established taxonomies, benchmarks, and playbooks for testing.
• Produce reports and datasets that clients can act on.

Connaissances

Red teaming experience

Curiosity and adversarial mindset

Structured approach

Strong communication skills

Adaptability

Description du poste

Obsidian is assembling a red team for AI model security, specifically focusing on identifying vulnerabilities and generating quality datasets. Candidates should have prior red teaming experience and a curiosity for probing AI systems. The role involves working with adversarial techniques to ensure AI safety.

Success in this position means highlighting risks AI models may overlook, enabling clients to strengthen their systems. This role offers an impactful position in a cutting-edge field, contributing to safer AI applications.