Researching the boundaries
of compute.
We build, stress-test, and evaluate systems — not just for performance, but for the world they operate in.
Model Architecture
Designing and training foundation models from the ground up. We explore novel architectures, optimize training pipelines, and push the limits of what models can learn and generalize.
Research · Development
Safety & Alignment
Rigorous red-teaming, adversarial testing, and alignment research. We probe systems for failure modes before they reach the real world, building trust through transparency.
Testing · Evaluation
Ethics & Governance
Intelligence without conscience is dangerous. We study the moral implications of autonomous systems and develop frameworks for responsible AI deployment at scale.
Policy · Philosophy
What we build
Software that
moves the world.
From internal tools to consumer platforms, we design and ship products that solve real problems — built on the same research that drives our AI work.
Pilot
Property management, reimagined. AI-driven maintenance, automated rent collection, and tenant communication — all in one platform.
Internal Tooling
Custom-built systems that power our operations — from model evaluation dashboards to deployment pipelines and data annotation workflows.
Community Contributions
We believe in building in the open. Select tools, libraries, and research artifacts are released publicly to push the ecosystem forward.