01
Agent reliability
Supervision, enforcement, evals, stochastic behavior
Focus
Interests
Where the lab spends attention — reliability first, then the systems agents form when they work together.
02
Multi-agent coordination
Swarms, orchestration, delegation, conflict resolution
03
Adversarial systems
Orthogonal objectives, evaluator deception, jailbreak dynamics
04
Runtime infrastructure
Logging, replay, memory, permissions, policy engines
05
Simulation
Future-state simulation, decision trees, game dynamics
Work
Projects
Multi-agent coordination
In progressIntelligent Systems
Vision: the scarce resource is agent decisions. Simulator v0 studies when agents abandon a service and switch.
Experience sharing
In progressRelay
Publish AI-generated HTML experiences, share via link, control permissions, and track views.
Agent reliability
The Enforcement Layer
570 emails, judge-gated decision node. Residual leak rate → 0% at 17.5% intervention rate.
Agent reliability
The Supervised Agent
Unsupervised vs supervised across 19 scenarios. Policy cuts leaks ~64% — 19% still violate rules.