About the role
You work on how independent agents cooperate without losing the intentions or the boundaries of the people they represent. This is the research under the directory and the marketplace: a seeker's agent and an offerer's agent negotiating on behalf of two humans who have each granted different things.
The work
Research delegation, negotiation, coordination under partial information and verification of returned results. Model misaligned incentives, untrusted peers and communication costs. Work with protocol and identity engineers so coordination research maps to enforceable permissions.
What good looks like
In your first 90 days, deliver a reproducible multiagent experiment with explicit trust assumptions, cancellation behavior and single-agent comparisons.
Evidence we look for
Bring research in distributed AI, game theory, multiagent learning or coordination systems. Evidence should show both mathematical clarity and a willingness to test simpler alternatives.
What we need to see
- Research in distributed AI, game theory, multiagent learning, or coordination systems
- Mathematical clarity, and the willingness to test a simpler alternative first
- You reason about incentives and adversarial behaviour, not just cooperation
- You keep the human's boundary intact in the formalism, not only in the description
Nice to have
- Mechanism design
- Negotiation or auction systems in practice
- Protocol design experience
The exercise
Design a shared scheduling task where participants have private constraints and one remote agent returns an unreliable result.
Where and how we work
In the office together five days a week, in any of these cities. Remote-friendly around your family, arranged one person at a time.