path-storage-policy

To organize project files across /home and /scratch, place sources, configurations, and logs under /home/$USER, while routing caches, datasets, and large artifacts to /scratch/$USER. This path storage policy ensures predictable placement and better quota management.

The best way to manage storage paths for large datasets and model weights is to route them to /scratch/$USER. This storage policy splits heavy artifacts from source code, keeping your /home directory clean and within quota limits.

You should use /scratch instead of /home when storing caches, datasets, model weights, and large artifacts. The storage policy recommends keeping /home/$USER for lightweight sources and configurations to maintain filesystem planning discipline.

You can enforce absolute path guidelines by applying a planning workflow that identifies read/write paths, classifies them by file type, and routes them to either /home or /scratch. This ensures consistent storage placement across all scripts and runtimes.

No specific dependencies are required to implement this storage policy. It provides a reusable layout blueprint and decision framework that can be applied directly to new projects needing storage discipline across /home and /scratch.