Researcher, Recursive Self-Improvement Safety (AI Safety)

TL;DR

Researcher, Recursive Self-Improvement Safety (AI Safety): Developing mitigation strategies and monitoring systems for recursive self-improvement to prevent loss of control in frontier AI systems with an accent on scalable oversight and automated auditing. Focus on designing rigorous monitorability tests, understanding model behavior science, and prototyping technical verification mechanisms.

Location: San Francisco

Salary: $295,000 – $445,000

Company

hirify.global is an AI research and deployment company dedicated to ensuring that general-purpose artificial intelligence benefits all of humanity.

What you will do

• Design and implement pre-deployment risk assessments and control measures for recursive self-improvement.
• Establish scalable oversight practices for monitoring model misbehavior in superhuman regimes.
• Develop automated auditing tools to detect severe model misalignments in production traffic.
• Conduct experiments in model behavior science to understand safety-relevant capabilities and risks.
• Prototype technical mechanisms for verifying compliance with future AI safety agreements.
• Translate research insights into established institutional practices and safety pipelines.

Requirements

• Exceptional technical execution skills.
• Strong strategic and research taste with the ability to prioritize in domains with weak feedback loops.
• Deep passion for mitigating risks associated with recursive self-improvement.
• Drive to perform work that positively impacts the future of AI development.
• Must be based in the US (as indicated by San Francisco location and regional legal compliance).

Nice to have

• Prior experience in ML research, AI alignment, or AI verification.

Culture & Benefits

• Commitment to equal opportunity and valuing diverse perspectives and experiences.
• Urgent, fast-paced work environment with far-reaching global and societal implications.
• Opportunity to push the boundaries of AI capabilities and safety.