# ADR 029: DVC

- HTML version: https://robbiepalmer.me/projects/homelab/adrs/008-dvc
- Project: Home Lab (https://robbiepalmer.me/projects/homelab.md)
- Status: Accepted
- Date: 2026-02-21
- Inherited from project: recipe-site (https://robbiepalmer.me/projects/recipe-site/adrs/029-dvc.md)

Applied to the lab: whichever machine runs batch jobs keeps local copies of
datasets on its own disk as a DVC remote, so those jobs read from a local
cache instead of re-pulling over the network each time. The same principles —
dataset hashes pinned to git commits, stage caching via `dvc repro`, and
metrics surfaced in PRs — govern how datasets are managed across the ML
pipelines (`ml-pipelines/recipe-parsing/`).

## Notes for Home Lab

Applied to the lab: whichever machine runs batch jobs keeps local copies of
datasets on its own disk as a DVC remote, so those jobs read from a local
cache instead of re-pulling over the network each time. The same principles —
dataset hashes pinned to git commits, stage caching via `dvc repro`, and
metrics surfaced in PRs — govern how datasets are managed across the ML
pipelines (`ml-pipelines/recipe-parsing/`).

---

Markdown index of this site: https://robbiepalmer.me/llms.txt