How we render 10K-asset scenes under 16ms
A look inside our distributed rendering pipeline and the scheduling tricks that keep frame latency flat as scenes scale.
Jul 2, 20268 min read
Ideas from the spatial AI frontier
Engineering deep-dives, product updates, and research from the team building the infrastructure for intelligent virtual worlds.
Latest
Recent writing
Long-form thinking on how we build, ship, and scale generative virtual environments.
Prompt-to-world: designing the generation API
Why we chose a declarative scene graph over raw diffusion output, and how it makes AI-generated environments editable.
Jun 18, 20266 min read
Spatial intelligence: context-aware object placement
The model architecture behind physically plausible layouts, and how we evaluate placement accuracy at scale.
Jun 3, 202611 min read
Building the infrastructure layer for virtual worlds
Our thesis on why the next generation of immersive experiences needs a dedicated spatial compute platform.
May 20, 20265 min read
Auto-scaling GPU fleets for bursty generation loads
How we keep costs low and throughput high when generation traffic spikes 20× during studio production sprints.
May 6, 20269 min read
Real-time collaboration inside generated environments
Multiplayer editing for 3D worlds — the CRDT model and conflict resolution that makes it feel instant.
Apr 22, 20267 min read