Public DGX/NVIDIA workstation notes without private topology.
Status: Public scaffold.
Release status: scaffolded.
License pending human review.
Human review required before release claims, license changes, external references, or publication of any real topology or implementation detail.
dgx-ai-workstation-notes is a public-safe documentation scaffold for explaining DGX/NVIDIA workstation workflow discipline without exposing the system behind it.
The repo preserves the DGX/NVIDIA lane in the Franzabner public surface as documentation architecture: how to classify artifacts, write runbooks, review tooling notes, and route future AI workstation work through human quality gates. It does not publish a real DGX map, a live lab topology, a training run, a hosted model, or a private runtime.
This scaffold is intended to show:
- workstation runbook discipline;
- generic NVIDIA tooling literacy;
- artifact status review before public claims;
- boundary rules for private topology, private corpora, private weights, private prompts, and private harnesses;
- relationship to
local-ai-lab-boundaryas the broader lab disclosure control; - relationship to future
fine-tuning-lab-notesas a downstream release-readiness lane; - proof routing back to the Franzabner public proof stack.
| Area | Public-safe proof |
|---|---|
| Workstation documentation | Shows how to write a workstation runbook without revealing real infrastructure. |
| NVIDIA lane readiness | Documents tooling categories, compatibility questions, and maintenance review shape at a high level. |
| Artifact discipline | Separates notes, templates, synthetic examples, and future review packets from releases or results. |
| HF caution | Keeps model, dataset, Space, report, benchmark, and eval language template-only unless separately approved. |
| Boundary control | Blocks private topology, hostnames, IP addresses, endpoints, credentials, corpora, weights, prompts, harnesses, and runtime internals. |
| File | Role |
|---|---|
STATUS.md |
Current scaffold status and non-claim posture. |
PUBLIC_BOUNDARY.md |
Public/private/sealed boundary for DGX and workstation notes. |
WORKSTATION_WORKFLOW.md |
Public-safe workflow model for documenting workstation work. |
NVIDIA_TOOLING_NOTES.md |
High-level NVIDIA tooling categories and review questions. |
PRIVATE_TOPOLOGY_POLICY.md |
Redaction and replacement rules for topology-bearing material. |
RUNBOOK_TEMPLATE.md |
Template for synthetic workstation runbooks. |
examples/synthetic-workstation-runbook.md |
Invented example packet showing the expected shape. |
CLAIMS.md |
Allowed and blocked claim language. |
REVIEW_CHECKLIST.md |
Human review checklist before external claims or real-detail publication. |
ROADMAP.md |
Public scaffold roadmap and review sequence. |
LICENSE_DECISION.md |
License pending human review. |
- Franzabner proof stack: https://github.com/Franzabner/franzabner-proof-stack
- Boundary companion: https://github.com/Franzabner/local-ai-lab-boundary
- HF release-readiness templates: https://github.com/Franzabner/hf-card-templates
- Future companion lane:
fine-tuning-lab-notes, planned/local-only until separately reviewed. - GitHub portfolio role: public-safe DGX/NVIDIA documentation scaffold after review.
- Upwork route: scoped workstation runbook, AI lab documentation, and artifact boundary documentation. Draft only until human-approved.
- Hugging Face route: process reference only. No model, dataset, Space, benchmark, eval result, hosted model, hosted dataset, or hosted artifact exists because this repo exists.
This is a public scaffold. No release, deployment, client, customer, revenue, production, model, dataset, Space, benchmark, eval result, accuracy, hosted artifact, or live service claim is made.
All examples are synthetic. Any future publication requires human review for topology, claims, license, Hugging Face posture, and proof-stack routing.