YOURSOTA — SAMPLE & PROCUREMENT BRIEF CHECKLIST
Human inputs, task instructions and engineering artifacts for AI post-training

This is a scoping checklist, not a dataset manifest, license or claim of verified rights. Values and evidence are established for the proposed purchase.

1. YOUR USE CASE
Model capability, task family, modalities, target languages, intended AI uses, volume, evaluation criteria and target delivery schedule.

2. CORPUS COVERAGE
Project count; file count; measured bytes; domains and disciplines; project types; source languages; original formats; native application versions; export compatibility; update coverage.

3. TASK STRUCTURE
Human-authored initial instructions; constraints; linked source references; follow-up instructions; retained correspondence; revision ordering; instruction-to-artifact links; validation/review records; documented acceptance status; missing stages.

4. QUALITY AND LINEAGE
Stable identifiers; source-file hashes; integrity checks; extraction limitations; missing files; duplicates; known quality issues; annotation methods; reviewer criteria; separation of original and derived materials; project-level train/evaluation holdouts.

5. PREPARATION SCOPE
Native source corpus versus prepared records; extraction; conversion; translation; anonymization/redaction; normalization; linking; task assembly; annotation; JSONL/Parquet or other agreed exports; expert review; acceptance criteria.

6. SOURCING AND RIGHTS
Source party; direct agreement or partner-assisted acquisition; relevant authorization; authority to grant the proposed license; training and evaluation scope; commercial model use; territory; duration; recipients; redistribution and derived datasets; third-party restrictions; confidentiality; treatment of trained models after license expiry.

7. COMMERCIAL AND DELIVERY TERMS
Representative sample; pilot; corpus license; optional expansion schedule; transfer method; access requirements; retention; delivery milestones; pricing; exclusivity if requested; warranties/remedies/liability terms to be negotiated.

CURRENT REVIEWED EXAMPLE
One 55 MB example archive from a growing corpus of more than 1 TB: 10 files across IFC (2), RVT (1), DWG (1) and DOCX (6). Source documents are in Russian. The technical brief includes tracked revisions and stated deliverable requirements. Corpus-wide volume, retained correspondence, complete task-to-artifact histories and completed validation evidence are assessed for the selected projects. Licensing scope and supporting authorization information are reviewed during diligence.

NEXT STEP
Submit your use case and diligence requirements at https://yoursota.ai/#contact to discuss sample availability and the proposed package.