Application Guide

How to Apply for Research Scientist, Manipulation Evaluations

at Apart Research

🏢 About Apart Research

Apart Research is a decentralized AI safety organization that uniquely combines cutting-edge technical research with a global community of researchers. They focus on making AI safety accessible and actionable, particularly through work relevant to emerging regulations like the EU AI Act. Working here offers the chance to contribute to high-impact, policy-relevant research in a flexible, remote-first environment.

About This Role

This role involves designing and building novel evaluations to detect harmful manipulation by AI systems, directly supporting EU AI Act enforcement. You'll develop risk scenarios, run evaluations on frontier models, and produce rigorous, defensible measurements that can inform regulation. It's a unique opportunity to shape how AI manipulation is assessed and mitigated at a systemic level.

💡 A Day in the Life

A typical day might start with a team standup to discuss progress on evaluation designs. You'll spend time reading recent literature on manipulation, then implement a risk scenario in code, running experiments against a frontier model. After analyzing results, you'll draft a section of a technical report or prepare findings for a regulatory briefing.

🚀 Application Tools

Generate Cover Letter

AI-powered, tailored for Apart Research

Interview Prep

Practice questions for this role

🎯 Who Apart Research Is Looking For

  • Has a strong background in AI safety or ML, with specific experience in designing evaluations or red-teaming AI systems for manipulation or deceptive behavior.
  • Is familiar with frontier models (e.g., GPT-4, Claude, Gemini) and understands how they can be used for manipulation, including social engineering, persuasion, or deception.
  • Can write clearly and technically for both regulators and researchers, with experience co-authoring technical reports or policy briefs.
  • Is proactive and self-directed, able to work remotely in a decentralized team, and comfortable with ambiguous, open-ended research problems.

📝 Tips for Applying to Apart Research

1

Highlight any past work on AI manipulation, persuasion, or deceptive alignment, even if from coursework or personal projects.

2

Include examples of evaluations or red-teaming you've designed, with details on methodology and results.

3

Show familiarity with the EU AI Act, especially provisions related to manipulation and systemic risk.

4

Tailor your resume to emphasize your ability to write for regulators—mention any policy or report writing experience.

5

In your cover letter, propose a concrete idea for a manipulation evaluation scenario you'd like to develop.

✉️ What to Emphasize in Your Cover Letter

['Your understanding of manipulation risks from frontier models and why they are a priority for AI safety.', 'Your experience designing rigorous evaluations or red-teaming exercises, with specific examples.', 'Your ability to translate technical findings into clear, actionable insights for non-technical regulators.', "Your alignment with Apart Research's mission of decentralized, community-driven AI safety research."]

Generate Cover Letter →

🔍 Research Before Applying

To stand out, make sure you've researched:

  • → Read Apart Research's published evaluations and blog posts on their website to understand their methodology and style.
  • → Study the EU AI Act's provisions on manipulation, particularly Article 5 (prohibited practices) and related risk categories.
  • → Review recent papers on AI manipulation, such as those from Anthropic or DeepMind on deceptive alignment and persuasion.
  • → Familiarize yourself with the concept of 'evaluation frameworks' used in AI safety (e.g., from ARC Evals or METR).
Visit Apart Research's Website →

💬 Prepare for These Interview Topics

Based on this role, you may be asked about:

1 Describe a manipulation evaluation you would design for a frontier model like GPT-4. What scenarios would you test?
2 How would you ensure your evaluation is both rigorous and defensible for regulatory purposes?
3 What are the key challenges in measuring AI manipulation, and how would you address them?
4 How do you stay updated on frontier model capabilities and emerging manipulation techniques?
5 Given Apart Research's decentralized structure, how would you collaborate with remote team members and the community?
Practice Interview Questions →

⚠️ Common Mistakes to Avoid

  • Applying without mentioning the EU AI Act or manipulation risks—shows lack of specific interest in the role.
  • Focusing only on general ML skills without demonstrating evaluation or red-teaming experience.
  • Submitting a generic cover letter that doesn't propose a concrete idea or show understanding of Apart Research's mission.

📅 Application Timeline

This position is open until filled. However, we recommend applying as soon as possible as roles at mission-driven organizations tend to fill quickly.

Typical hiring timeline:

1

Application Review

1-2 weeks

2

Initial Screening

Phone call or written assessment

3

Interviews

1-2 rounds, usually virtual

✓

Offer

Congratulations!

Ready to Apply?

Good luck with your application to Apart Research!

← Back to Job Listing Apply Now →