How to apply for Research Scientist

About AI Verification and Evaluation Research Institute

AVERI is a nonprofit focused on ensuring powerful AI systems are rigorously audited by independent third parties, filling a critical gap in AI safety and security. Unlike many AI organizations, it prioritizes external verification and policy-relevant research, making it an ideal place for those who want to hold AI developers accountable. Working here means contributing to a growing field that blends technical rigor with real-world impact on AI governance.

About the role

As a Research Scientist at AVERI, you'll design and run evaluations of frontier AI systems to assess capabilities, risks, safety, and alignment, while developing audit methods and protocols. You'll build reproducible pipelines and produce reports with actionable recommendations for developers and policymakers, directly shaping how AI companies are held accountable. This role is impactful because your research will inform third-party audits and potentially influence AI policy and industry practices.

A typical day

A typical day might involve designing a new evaluation protocol for a frontier AI model, writing code to run experiments, and analyzing results to identify safety or compliance issues. You might also collaborate with engineers to refine pipelines, draft a section of an audit report, or meet with policy experts to discuss implications of your findings. Expect a mix of hands-on technical work and strategic discussions about how to improve AI accountability.

Who AI Verification and Evaluation Research Institute is looking for

• Experience designing and conducting empirical evaluations of AI systems, especially large language models or other frontier models, with a focus on safety, robustness, or alignment.
• Strong background in developing audit methodologies or protocols for assessing model behavior, including red-teaming, adversarial testing, or policy compliance checks.
• Proficiency in building reproducible experimental pipelines and tools, with solid software engineering skills (e.g., Python, ML frameworks, data versioning).
• Track record of publishing research (papers, white papers, or tools) and communicating findings to diverse stakeholders, including technical and policy audiences.

Tips for this application

• Highlight any experience you have with AI auditing, evaluations, or red-teaming in your resume and cover letter—AVERI is specifically looking for people who can design and run assessments, not just build models.
• Demonstrate familiarity with AVERI’s mission and existing work: mention specific reports, tools, or publications from their website, and explain how your skills align with their audit-focused approach.
• Provide concrete examples of reproducible pipelines or tools you’ve built for evaluating AI systems, including links to GitHub repos or published artifacts.
• Emphasize your ability to translate technical findings into actionable recommendations for policymakers and developers, as this is a key output of the role.
• If you have experience working with AI companies or on third-party audits, explicitly describe it—AVERI values candidates who understand the practical challenges of auditing frontier AI.

What to cover in your cover letter

['Your specific experience in designing and running evaluations of AI systems, particularly for safety, security, or alignment, and how it prepares you to develop audit methods at AVERI.', 'Your ability to build reproducible experimental pipelines and produce clear, actionable reports, with examples of past work that had real-world impact.', 'Your alignment with AVERI’s nonprofit mission and understanding of why third-party audits are essential for AI safety and policy.', 'Your collaborative skills and experience working across disciplines (e.g., with engineers, policy experts) to publish research and tools.']

Research before applying

• Read AVERI’s published reports, white papers, and blog posts to understand their audit methodology and focus areas (e.g., specific risks like deception, bias, or security).
• Explore their partnerships or collaborations with AI companies or policymakers to gauge their influence and approach.
• Look into the backgrounds of their team members, especially researchers and leadership, to understand their expertise and research interests.
• Familiarize yourself with current AI audit frameworks (e.g., NIST AI RMF, EU AI Act) and how AVERI’s work fits into the broader ecosystem.

Likely interview topics

Based on the job description, expect questions about:

• How would you design an evaluation to assess a frontier AI model’s robustness to adversarial attacks or its compliance with safety policies?
• Describe a time you developed an audit protocol or evaluation pipeline. What were the key challenges and how did you ensure reproducibility?
• How do you balance technical rigor with the need for actionable recommendations that policymakers and developers can implement?
• What do you see as the biggest gaps in current AI auditing practices, and how would you address them at AVERI?
• Given AVERI’s nonprofit status, how would you handle potential conflicts of interest or pressure from AI companies during an audit?

Common mistakes to avoid

• Focusing only on model development or traditional ML research without addressing evaluation, auditing, or safety—this role is about assessing AI systems, not building them.
• Being vague about your experience with reproducibility or tool-building; AVERI needs concrete examples of pipelines you’ve created and maintained.
• Ignoring the policy and governance context of AI auditing; showing no awareness of how your research could inform regulation or industry standards is a major turn-off.