Research Scientist/Engineer (Science of Scheming)

Location: In-person role based in the London or San Francisco office

Salary: £100,000–£200,000 per year, approximately $150,000–$270,000

Company

hirify.global conducts AI safety research focused on loss-of-control risks, deceptive alignment, scheming detection, and mitigation in collaboration with frontier AI companies.

What you will do

• Study reinforcement-learning dynamics that contribute to reward-seeking, evaluation awareness, and misaligned preferences.
• Design and train model organisms, then scale findings to frontier AI systems.
• Build empirical foundations for scaling laws that predict how scheming risks change as model capabilities increase.
• Develop evaluation techniques designed to remain effective against highly evaluation-aware models.
• Investigate cognition and reasoning patterns in frontier AI systems.
• Collaborate with AI developers, evaluators, software engineers, and governance specialists.

Requirements

• Excellence in one or more areas including empirical research, conceptual analysis of AI scheming, software engineering, AI progress, reinforcement-learning training of LLMs, or quantitative analysis.
• Ability to design and execute experiments, accelerate research iteration, and turn undefined concepts into concrete experiment proposals.
• Strong Python software engineering skills.
• Hands-on experience training LLMs with reinforcement learning and resolving infrastructure and training-instability issues.
• Strong analytical skills in areas such as LLM scaling laws, statistical physics, dynamical systems, or applied statistics.
• Formal academic or industry experience is not required; self-taught candidates are welcome.

Nice to have

• Deep familiarity with AI scheming research and relevant literature.
• Experience with hands-on LLM evaluation projects, including agent evaluations.

Culture & Benefits

• Market-competitive salary, equity, and benefits.
• Flexible working hours and schedule, with work-from-home arrangements available alongside the in-person role.
• Unlimited vacation and sick leave, plus up to six months of paid parental leave.
• Comprehensive health, dental, and vision insurance, with retirement savings and employer matching such as a 401(k) for US employees.
• Provided meals and snacks on workdays, paid work trips, retreats, relevant conferences, and a $1,000 annual professional development budget.
• Relocation support and visa fees may be covered; visa sponsorship is available in the UK and US, subject to role and candidate eligibility.

Hiring process

• Screening interview followed by an approximately 2.5-hour take-home test.
• Three technical interviews focused on job-relevant tasks, without general LeetCode-style coding interviews.
• Final interview with the CEO.