Jailbreaking Lead (Red Team)

About the Company

FAR.AI is a non-profit AI research institute dedicated to ensuring advanced AI is safe and beneficial for everyone. Since our founding in July 2022, we've grown quickly to 45+ staff, producing over 40 influential academic papers and establishing leading AI Safety events.

Responsibilities

• Personally develop universal and near-universal jailbreaks against frontier closed- and open-weight models in domains like CBRNE, cyber, and agentic security
• Systematically dismantle defense-in-depth stacks including input filters, model-level refusal, and reasoning monitors
• Invent new attack classes and monitor state-of-the-art methods from literature
• Mentor ICs on attack craft and review major red-teaming deliverables for technical quality
• Work directly with frontier labs and government agencies to ensure findings lead to real mitigations
• Shape the jailbreaking research agenda and stress-test novel affordances like multimodal and long context

Requirements

• Proven track record of finding non-obvious, high-severity vulnerabilities in frontier AI systems
• Deep, hands-on jailbreaking experience with modern frontier models and layered defenses
• Experience with black-box optimization, multimodal attacks, or agentic red-teaming
• Deep understanding of LLM architectures, training processes, and failure modes
• Strong technical background in AI, adversarial ML, security, CS, or math
• Ability to thrive in fast-moving, ambiguous environments

Preferred Qualifications

• Prior collaboration with AI labs, security teams, or government safety institutes
• Track record in top CTF teams or offensive security research
• Published work in AI safety, security, or robustness
• Experience mentoring technical ICs or leading small technical teams