For data owners
Preparing engineering data for licensing: a checklist for owners
Before you approach buyers, these steps make a review faster and a license easier to agree.
Companies often hold years of engineering history — repositories, issue trackers, code reviews — that could be valuable for AI development. Turning that history into a licensable dataset is mostly a matter of preparation and paperwork.
Before you start
- Confirm who owns the code: check employment and contractor agreements for IP assignment.
- List client work separately; client contracts frequently restrict reuse.
- Identify vendored and copied third-party code and its licenses.
- Decide which uses you are comfortable licensing, and which you are not.
Technical preparation
- Scan for secrets and credentials, and rotate anything found — removal from data is not enough.
- Remove or pseudonymise personal data such as names and email addresses where required.
- Document the schema and how records link together.
- Make sure a small, representative part of the repository can serve as a sample: it is taken from the repository you connect, never sent by email.
Set expectations
Licensing outcomes depend on buyer demand, data characteristics, and the terms both sides can agree. A good preparation process makes your data easier to evaluate; it cannot guarantee a sale or a price.