AI Safety Expert – Red Team

Join mercor as an AI Safety Expert - Red Team, working on-site in Pakistan to strengthen the safety and reliability of conversational AI systems. This full-time role offers the opportunity to identify and document vulnerabilities in AI models before they reach production, contributing directly to responsible AI development.

Position Summary

We're seeking experienced red teamers to conduct adversarial testing of conversational AI models and agents. Your work will involve systematically probing AI systems for security weaknesses, misuse vulnerabilities, and harmful outputs. By combining structured testing methodologies with creative problem-solving, you'll help identify risks that need to be addressed before AI reaches real-world deployment. This role requires fluency in both English and Urdu, enabling you to test and evaluate AI responses across linguistic and cultural contexts.

Day-to-Day Responsibilities

• Conduct red team exercises on conversational AI models and agents, uncovering jailbreaks, prompt injection vulnerabilities, and potential misuse scenarios.
• Produce high-quality annotated data by documenting failures, categorising vulnerabilities, and identifying systemic risks in AI behaviour.
• Apply consistent testing frameworks by working within established taxonomies, benchmarks, and documented playbooks to ensure rigorous coverage.
• Create reproducible documentation including detailed reports, datasets, and documented attack methodologies that clients can understand and act upon.
• Evaluate AI responses on sensitive subjects including bias, misinformation, and harmful content, with optional involvement in higher-risk assessment areas.

What We're Looking For

Essential

• Native or near-native fluency in both English and Urdu.
• Demonstrated experience in red teaming, whether through AI adversarial testing, cybersecurity work, or socio-technical security research.
• Proven ability to communicate security findings clearly to both technical specialists and non-technical decision-makers.

Valued experience

• Background in adversarial machine learning, cybersecurity, or socio-technical threat analysis.
• Skills in creative adversarial thinking gained through psychology, performance, creative writing, or unconventional problem-solving approaches.

Working at mercor

mercor connects exceptional technical and creative talent with leading artificial intelligence research organisations globally. The company partners with frontier AI labs to ensure their systems are thoroughly tested and made safer before deployment. If you're passionate about AI safety and want to work on genuinely challenging problems with real-world impact, this opportunity aligns with that mission.

How to Apply

To apply, complete your application directly on this page, or you'll be redirected to the employer's application platform to finish submitting there.