2-node NVIDIA GB10 (DGX Spark)

Self-hosted GPU cluster for local inference.

Connected concepts

Local-first inference, vLLM / PagedAttention