Evidence First
We follow the data, publish our reasoning internally, and change our minds in public.

CAREERS
We are a small, deliberate team building general intelligence systems for the long term. We hire rarely, and we hire well.
WHAT WE LOOK FOR
DISCIPLINES
We organize around problems, not titles. These are the disciplines where we most often look for new colleagues.
Framing the questions that matter — in world modeling, representation learning, and reasoning — and designing the pre-registered, ablation-driven experiments that settle them.
Turning ideas into systems that actually run: learned world models, planners, evaluation harnesses. The craft of making experiments fast, honest, and repeatable.
Building the measured, dependable substrate a long-horizon research program stands on: data pipelines, reproducible training, and instrumentation that fails loudly and recovers gracefully.
Keeping a focused organization running well — from partnerships and communication to the quiet operational excellence that lets researchers do their best work.
OPEN APPLICATION
We don't post conventional job listings. If our mission resonates and you believe your work speaks for itself, write to us — tell us what you've built, what you want to build next, and why here.
INTRODUCE YOURSELF