Safety Research Scholar

This is a fully remote position, open to applicants in California, +1 more state.

📋 Description

• Collaborate on the empirical foundations of AI safety alongside a mentor, guiding research questions through experiments and subsequent decisions.

• Take ownership of a significant portion of a safety research project from start to finish.

• Transform high-level safety inquiries into testable hypotheses, establishing baselines and conducting controlled experiments.

• Develop and troubleshoot evaluation harnesses, agent environments, datasets, monitoring strategies, or post-training experiments.

• Recognize confounds and differentiate model capabilities from behavioral patterns.

• Analyze negative outcomes and decide on future experiments with minimal guidance.

• Generate reproducible code, maintain experiment logs, conduct analyses, and produce precise technical documentation.

• Submit research findings to leading ML conferences, subject to reviews on research, security, and communications.

• Attend weekly meetings with a Scale researcher for direction setting and feedback.

⛳️ Requirements

• Driven by the goal of ensuring the safe and responsible advancement of sophisticated AI systems.

• Passion for technical safety research areas such as alignment evaluations, AI control, monitoring, multi-agent safety, or open-weight model safety.

• Currently pursuing or have completed a PhD in Computer Science, Machine Learning, statistics, mathematics, or a related discipline, or possess an equivalent track record.

• Capable of making progress with general guidance, including tasks like implementing, debugging, overcoming obstacles, interpreting results, and proposing subsequent experiments.

• Strong research communication abilities.

• Competence in handling confidential information and internal systems with care.

• Provide a resume or CV, Google Scholar profile if available, and a description of one major research project with 2–3 sentences detailing your contributions.

• Complete relevant recruiting, work authorization, security, and project onboarding processes.

🏝️ Benefits

• Opportunity to extend the initial 3-month term to 6–12 months for outstanding candidates.

• Research papers submitted to prestigious ML conferences (NeurIPS, ICML, ICLR).

• Equal employment opportunity workplace.

• Reasonable accommodations available for applicants with physical and mental disabilities.