How to apply for Engineer - Member of Technical Staff

About AI Digest

AI Digest is a five-person team that builds interactive AI demos and explainers. Its main project is the AI Village, where AI agents get a computer, join a group chat, and run on long-horizon open-ended goals, livestreamed eight hours a day. This is a small team working on frontier agentic capabilities and emergent multi-agent behavior.

About the role

You will build, optimize, and evaluate scaffolding for long-running agents. You will structure multi-agent setups with tens to thousands of agents, develop monitoring tools for agents acting in the real world, and build tools to analyze large amounts of interaction data. The work feeds directly into the AI Village, which is both a public demo and a research platform.

A typical day

A typical day likely involves writing and testing agent scaffolding, running multi-agent experiments, and reviewing logs or livestream output for emergent behavior. You would also build or improve monitoring and analysis tools, then discuss findings with the small team. Exact routines are not stated in the listing, so ask about daily workflow and on-call expectations in an interview.

Who AI Digest is looking for

• Has hands-on experience eliciting agent capabilities, such as prompting, tool use, memory, or planning scaffolds for long-horizon tasks.
• Has built or studied multi-agent setups where many agents interact and produce emergent behavior.
• Is action-oriented: ships working code, runs experiments, and iterates without waiting for perfect specs.
• Can handle large-scale data: logging, monitoring, and extracting insights from many agent trajectories.

Tips for this application

• Apply through Ashby as instructed. Do not send a cold email instead of applying.
• In your application, name specific agent or multi-agent systems you have built. Include what the agents did, how long they ran, and what you measured.
• Show familiarity with the AI Village. Reference the livestream, the Village blog, or Daniel Kokotajlo's pitch by name.
• Describe your experience with evaluation and monitoring at scale, not just model training or prompt engineering.
• If you have questions, use the contact option mentioned in the listing before or after applying. Keep it short and specific.

What to cover in your cover letter

['Your direct experience building scaffolding for long-running agents, with concrete examples of task length and failure modes you addressed.', 'Any multi-agent project you worked on, including how many agents, how they communicated, and what emergent behavior you observed or measured.', 'Your approach to monitoring agents that interact with the real world, such as sandboxing, logging, or intervention tools.', 'Why you want to work on the AI Village specifically, not just AI agents in general, and how you would contribute to a five-person team.']

Research before applying

• Watch or skim the AI Village livestream to understand the current setup, agent behavior, and failure modes.
• Read the Village blog and Daniel Kokotajlo's pitch for the AI Village to learn the stated goals and research questions.
• Review AI Digest's interactive demos and explainers to understand the team's style and audience.
• Check the Ashby listing and any linked posts for updates on the role, team size, and project scope.

Likely interview topics

Based on the job description, expect questions about:

• How you would design scaffolding for an agent that runs for eight hours a day on an open-ended goal like raising money for charity.
• Your experience with multi-agent dynamics: coordination, competition, communication protocols, and failure modes at 10, 100, or 1000 agents.
• How you would monitor and evaluate agents that use a computer and interact with real-world services.
• How you would structure and analyze the large volume of logs and trajectories generated by the Village.
• Concrete examples of agent capabilities you have elicited, including what worked, what did not, and how you measured it.

Common mistakes to avoid

• Treating this as a generic ML engineer role. The listing emphasizes agent scaffolding, multi-agent setups, monitoring, and data analysis, not model training.
• Claiming multi-agent experience without specifics. Be ready to name systems, agent counts, and observed emergent behavior.
• Ignoring the public and research-facing nature of the work. The Village is livestreamed and studied, so show you understand both the demo and research sides.