Byte-Identical
Verification at Scale
Reproducibility is usually a convention: a README says which data produced a result, and everyone trusts it. Here it is a checksum pass. A full drive copy, with that file count verified byte-for-byte against a GPG-signed SHA-256 manifest. Home lab, code available on request.
Why Convention Is Not Enough
A large file migration is only as trustworthy as the answer to "did every byte survive the copy." When that answer lives in a file count or a progress bar, it drifts silently: a file truncates, a copy stalls, and nobody notices until something reads back wrong. A checksum manifest removes the ambiguity. Every file's SHA-256 digest is recorded before the copy and checked again after, so there is no such thing as a silent divergence.
This case is that discipline run at full-corpus scale on a home-lab drive migration: not a claim of a data-versioning platform, just the checksum-verify mechanism, GPG-signed and re-run until the result is 0 divergent files.
The Two Properties
Why This Matters For A Team
A platform or data team migrating a large file set needs to answer "did anything silently corrupt" without hand-waving. A GPG-signed checksum manifest makes that answer a verify command instead of a promise. The manifest signs rather than seals, so it never becomes a lockout risk for the owner. Reproducibility becomes an exit code.
reproducibility discipline - Remote - open to mid-level and senior IC roles
