ML Research Team Lead

About the company

An international nonprofit focused on AI safety. The team develops technologies that improve the robustness and reliability of AI systems. They raise awareness among policymakers about AI risks and aim to accelerate progress in technical safety.

They are currently looking for ML Research Team Lead — a seasoned researcher with strong expertise who can define priorities, shape research directions, and support a team of researchers in achieving meaningful results in the area of AI safety.

What You’ll Do:

• Set Research Direction: Work with the current research backlog and propose new project ideas.
• Design and Execute Projects: Formulate hypotheses, design experiments, and drive projects from idea to publication.
• Team Leadership: Manage a team of researchers, provide technical mentorship, support experiment planning, and ensure quality and reproducibility.
• Communication: Present results to both technical and general audiences, write papers and blog posts, and collaborate with partners (such as RAND, AISI, and others).

Who We’re Looking For:

• 6+ years of technical experience, including research projects and leadership roles.
• English proficiency at C1 level or higher.
• Deep understanding of LLMs — their capabilities, limitations, and evaluation methods (prompting, CoT/ToT, RAG, tool use, agents).
• Expert-level Python skills and the ability to set a high technical bar for the team.
• Strategic mindset with the ability to identify high-value directions in a fast-changing field.
• Proven team management experience.
• Strong communication skills.

Bonus Skills:

Experience with research planning, AI safety evaluation methods (e.g. Anthropic, Apollo, METR), red-teaming techniques, reproducible science, and cross-functional collaboration with engineering and comms teams.

What They Offer:

• Full-time remote work from any location in a timezone close to Central European Time.
• A chance to work on cutting-edge technologies in AI and AI safety — where science meets practical impact.
• Fast-paced research-to-impact cycle: the team shares its findings with the world through papers, blogs, interactive demos, and more. Results are also regularly used by international partners in briefings and deployment.