Multi-runtime GPU + remote inference as a supervised actor system on atomr. Single-crate rollup that re-exports per-runtime backends behind feature flags.
No-std WASM kernel for 256-tile coherence gate fabric
Parallel SSH command execution tool for cluster management