Remote AI Safety Red Team Specialist

Mercor; Lisboa; Presencial; EUR 40 000 - 65 000; Tempo integral

Resumo da oferta

Mercor is assembling a red team to test AI models with adversarial inputs, surface vulnerabilities, and generate data to improve safety for customers. The role involves reviewing outputs on sensitive topics and collaborating with stakeholders across projects.

Participation in high-sensitivity tasks is supported with guidelines and wellness resources. You will contribute by red-teaming models, generating high-quality data, following taxonomies, and documenting reproducible reports and attack

Qualificações

• Prior red teaming experience in AI adversarial work, cybersecurity, or socio-technical probing.
• Must be curious and adversarial, able to push systems to breaking points.
• Structured approach using frameworks or benchmarks, not ad-hoc hacks.
• Clear communication of risks to technical and non-technical stakeholders.
• Able to adapt across projects and customers.

Responsabilidades

• Red team conversational AI models and agents: jailbreaks, prompt injections, misuse cases, bias exploitation, multi-turn manipulation.
• Generate high-quality human data: annotate failures, classify vulnerabilities, and flag systemic risks.
• Apply structure: follow taxonomies, benchmarks, and playbooks to keep testing consistent.
• Document reproducibly: produce reports, datasets, and attack cases customers can act on.

Conhecimentos

Red teaming experience

Adversarial mindset

Structured thinking

Clear communication

Adaptability

Descrição da oferta de emprego

Mercor is assembling a red team to test AI models with adversarial inputs, surface vulnerabilities, and generate data to improve safety for customers. The role involves reviewing outputs on sensitive topics and collaborating with stakeholders across projects.

Participation in high-sensitivity tasks is supported with guidelines and wellness resources. You will contribute by red-teaming models, generating high-quality data, following taxonomies, and documenting reproducible reports and attack