Principled Agents is hiring a Researcher and a Research Assistant for a project developing methods to have AI consider uncertainty in their learned goals.
These are contract roles, funded through a grant from the UK’s AI Security Institute, and administered by Principles of Intelligence. We are interested in potentially transitioning successful contractors to long-term positions, but cannot yet make any commitments.
Responsibilities
- Plan and run experiments training and evaluating language models, using Python libraries such as PyTorch
- Collaborate on developing mathematical models of the incentives faced by AI
- Analyze experimental results and assist in the writing of research papers
Requirements
- Experience conducting technical AI safety research
- Strong mathematical modelling ability
- Understanding of machine learning
- Preference for applicants with a graduate degree in a related field (CS, Mathematics, Economics, Physics) and/or previous employment in AI
- Clear technical writing skills
Hiring Process
Our hiring process consists of the following steps. Applications are processed on a rolling basis, and we may invite applicants to the next stage or even make offers before the end of the application deadline.
1. Application:Â Submit an application through the form linked above by May 3rd, 2026 at 11:59pm (anywhere on Earth).
2. Interview:Â Attend two 30-minute remote interviews, one focused on background/fit and one focused on evaluating technical skills. These two interviews may be combined.
3. References:Â Provide references who can speak to your qualifications.
4. Offer:Â Successful candidates will be made an offer to begin working at an agreed-upon date.
