Skip to content
Karmyug

Already sitting on data an AI team needs?

List it on the Karmyug catalogue. You keep ownership; we handle compliance review, licensing and buyers.

What we can list

  • Speech, image, video, text and multimodal data
  • Clear provenance and a consent record you can produce
  • Rights you actually hold, including for commercial licensing
  • A specification: volume, languages, capture method, demographics
  • Annotations, if any, with the guideline used to produce them

What we will not list

  • Data scraped from platforms whose terms forbid it
  • Biometric data collected without specific, informed consent
  • Personal data you cannot show a lawful basis for
  • Datasets you licensed from someone else and cannot sublicense

This is not caution for its own sake. A dataset with unclear provenance is a liability that surfaces during the buyer's diligence, and it damages both of us.

How listing works

Five steps, with the honest conversation at step three.

  1. 01

    Submit a description

    Volume, modality, languages, how it was collected and what rights you hold.

  2. 02

    Compliance review

    We check provenance, consent and rights. This is the step most submissions fail.

  3. 03

    Quality sample

    We audit a sample and tell you honestly what the set is worth and why.

  4. 04

    Listing and licensing

    We list it, handle buyer enquiries, licensing and payment collection.

  5. 05

    Revenue share

    You keep ownership. Revenue is split per the agreement, paid quarterly.

Questions from data owners

Do I lose ownership of my data?

No. You retain ownership and grant us the right to sublicense it under agreed terms. You can withdraw future licensing at any time, subject to licences already issued.

What revenue share do you offer?

It depends on exclusivity, how much preparation the set needs and how differentiated it is. We give you a number after the quality audit, not before.

Why would you reject a dataset?

Almost always provenance. If the consent record cannot be produced per file, or the rights were never yours to grant, we cannot list it and neither should anyone else.

Can you prepare the data for listing?

Yes. Cleaning, format conversion, de-identification and annotation can be done as paid work or offset against revenue share.

Tell us what you have

Volume, modality, languages, how it was collected and what rights you hold. We will tell you quickly whether it is listable.