Week 23 · 1–7 Jun 2026
The self-hosted stack moved onto an NVIDIA DGX Spark, serving language models locally behind a model gateway.
· Written
- PRs opened
- 6
- PRs merged
- 5
- PRs reviewed
- 0
- Issues resolved
- 0
- Active repositories
- 3 of 5
Counts cover 1–7 Jun 2026 (UTC) across the project's tracked repositories. Pull request figures count only pull requests opened during these dates; merged, closed and reviewed mean that also happened within them, so one can count in more than one category. They count pull requests of any size, including many prepared by AI coding agents and some fully automated ones. Reviewed means a submitted review by someone other than the author, not approval.
The self-hosted stack moved onto real hardware this week. On an NVIDIA DGX Spark it began serving language models locally, behind a model gateway, with dashboards to watch them and HTTPS for the workflow engine. Alongside, the core principles and the writing instructions that cite them were brought into line, and the sixth post in the blog series was produced in full. Written retrospectively from the project's history.
What we worked on
- Reference stack. The self-hosted stack began running on an NVIDIA DGX Spark, with local model serving behind a model gateway, dashboards that monitor model use and HTTPS for the workflow engine.
- Agents and runtime. We documented a personal AI setup for a laptop that connects to the model gateway.
- Platform specification. We wrote an install runbook for the local AI stack, and brought the core principles and the documents that cite them into line.
- Content. We produced the full set of assets for the sixth post in the blog series and set up a production framework for the rest.
01-Jun-2026
- Platform specification. We wrote an install runbook for the local AI stack on the NVIDIA DGX Spark.
- Reference stack. The self-hosted stack gained its configuration and services for the DGX Spark.
02-Jun-2026
- Reference stack. We added a runbook and configuration for serving language models locally on the DGX Spark.
03-Jun-2026
- Reference stack. Local model serving gained monitoring dashboards, and the workflow engine moved to HTTPS.
04-Jun-2026
- Platform specification. The core principles and the writing instructions that cite them were brought into line.
- Content. We produced the full set of assets for the sixth post in the blog series and set up a production framework for the rest.
05-Jun-2026
- Reference stack. A model gateway with its own database now sits in front of local model serving.
- Agents and runtime. We documented a personal AI setup for a laptop that connects to the model gateway.
06-Jun-2026
- Reference stack. The model gateway's settings moved into its database.
Still in progress
- Routing every model, local and on the laptop, through one gateway is being set up.
- The next posts in the blog series are being prepared.
Looking ahead
Next we turn to bringing every model behind the one gateway, with tracing. The blog series continues alongside.