Safety Research Scholar (AI Safety)

Location: Remote within the United States, including California and New York

Compensation: $100/hour

Company

Scale AI develops reliable AI systems, high-quality data, and full-stack technologies for leading models, enterprises, and governments.

What you will do

• Own an AI safety research project from defining hypotheses and baselines through controlled experiments and research outputs.
• Implement and debug evaluation harnesses, agent environments, datasets, monitoring methods, and post-training experiments.
• Analyze model behavior, identify confounds, interpret negative results, and propose subsequent experiments with high-level guidance.
• Produce reproducible code, experiment logs, technical analyses, and precise research writing.
• Submit research papers to top ML conferences including NeurIPS, ICML, and ICLR, subject to research, security, and communications review.

Requirements

• Motivation to support the safe and responsible development of advanced AI systems.
• Interest in technical AI safety research, including alignment evaluations, AI control, monitoring, multi-agent safety, or open-weight model safety.
• Pursuing or having completed a PhD in computer science, machine learning, statistics, mathematics, or a related field; researchers with an equivalent track record are also encouraged to apply.
• Ability to make progress with high-level guidance, including implementing, debugging, interpreting results, getting unstuck, and proposing next experiments.
• Clear research communication skills and responsible handling of confidential information and internal systems.

Culture & Benefits

• Part-time internship structure at 20 hours per week.
• Initial term of 3 months, with an option to extend to 6–12 months for exceptional candidates.
• Weekly meetings with a Scale researcher who provides direction and feedback.
• Inclusive and equal opportunity workplace with reasonable accommodations available for applicants with disabilities.

Hiring process

• Submit a resume or CV, a Google Scholar profile if available, and one substantially owned research project with 2–3 sentences describing your contribution.
• Discuss a past research project, including attempted approaches, failures, and lessons learned.
• Complete a research-design exercise and propose an approach to a research problem.
• Successful candidates complete applicable recruiting, work-authorization, security, and project-onboarding steps. Candidates must wait 90 days before reconsideration for the same role.