Bjorn Hunter Professional Corporation

Custom, self-hosted AI engineering — Edmonton, Canada

A private three-node NVIDIA DGX Spark supercluster, built and operated end to end: hardware, fabric, serving stack, evaluation, and the local agents that run on it.

The cluster

Compute3 × NVIDIA DGX Spark (GB10 Grace-Blackwell), 128 GB unified memory each
Fabricdirect 200 Gb/s QSFP per node pair — full triangle, no switch
Flagship inferenceDeepSeek-V4.1-Flash-1017, EXL3, tensor-parallel across 3 nodes
Context1,048,576 tokens · 4 concurrent sequences · 2.2 GiB KV cache
Throughput53.9 tok/s single stream · 137.6 tok/s aggregate, 4-stream burst
Long context443,223-token needle retrieved correctly in 87.1 s

Alongside the flagship model the cluster runs local vision, embedding and reranking, speech synthesis, and image/video generation — each scheduled into its own window so they never contend with inference.

Published work

How I work

Measured, not claimed. Every number here was produced on these machines under a named configuration, with the result files published alongside it — including the superseded runs and the weak spots.