How to apply for Research Scientist - Member of Technical Staff

About AI Digest

AI Digest is a five-person team that builds interactive AI demos and explainers about the future of AI. Their main project, the AI Village, gives AI agents a computer, puts them in a group chat, and runs them on long-term goals like raising money for charity, all livestreamed eight hours a day.

About the role

You will build and study the AI Village, structuring multi-agent setups with tens to thousands of agents and turning large amounts of village data into rigorous research. You will also run new experiments to test hypotheses using a large compute budget. The goal is to demonstrate frontier long-horizon agentic capabilities and enable research into model behaviour and emergent multi-agent dynamics.

A typical day

A typical day might involve reviewing livestream data from the AI Village, designing a new multi-agent experiment, and running analyses to test a hypothesis about emergent behaviour. You would also spend time discussing research directions with the small team and iterating on setups for hundreds or thousands of agents. Since the role is remote and the team is five people, expect a lot of independent work and direct communication.

Who AI Digest is looking for

• Has a strong sense of which research questions are interesting and important, and can quickly design and execute great analyses and experiments.
• Gets to the bottom of things and can work with vast amounts of data from multi-agent interactions.
• Is comfortable structuring multi-agent setups where hundreds or thousands of agents interact and produce emergent patterns.
• Can run new experiments with a large compute budget and turn results into insightful, rigorous research.

Tips for this application

• Apply through Ashby as instructed in the job post. Do not send applications by email unless you have a question.
• Reference the AI Village specifically. Mention Daniel Kokotajlo's pitch and the Village blog if you have read them.
• Describe concrete examples of past work with multi-agent systems, agentic capabilities, or large-scale behavioural data analysis.
• Explain how you would structure an experiment with 100s or 1000s of agents to test a hypothesis about emergent behaviour.
• If you have questions, use the contact method mentioned in the post instead of guessing about team or culture.

What to cover in your cover letter

['Your ability to identify interesting research questions about multi-agent dynamics and long-horizon agentic capabilities.', 'Specific experience designing and executing analyses or experiments on large behavioural datasets.', 'Concrete ideas for how you would study or extend the AI Village, such as new multi-agent setups or experiments.', "Your familiarity with AI Digest's public work, including the AI Village livestream and blog."]

Research before applying

• Read Daniel Kokotajlo's pitch for the AI Village to understand the project's goals and framing.
• Read the Village blog for updates, findings, and examples of past experiments.
• Watch part of the AI Village livestream to see how the agents interact and what data is generated.
• Review AI Digest's interactive AI demos and explainers to understand their approach to making sense of the future.

Likely interview topics

Based on the job description, expect questions about:

• How would you structure a multi-agent setup with 1000 agents to produce emergent behavioural patterns?
• Given the AI Village data, what research question would you ask first and how would you analyze it?
• Describe an experiment you would run to test a hypothesis about long-horizon agentic capabilities with a large compute budget.
• What do you think are the most interesting or important open questions in multi-agent dynamics?
• How would you turn a vast amount of village data into rigorous, insightful research?

Common mistakes to avoid

• Submitting a generic research statement that does not mention the AI Village or multi-agent systems.
• Focusing only on model training or ML engineering without showing interest in behavioural research or emergent dynamics.
• Ignoring the application instructions, such as applying through a different channel or failing to ask questions when unclear.