unthinkable
Case study · IndiaAI, Ministry of IT, with NeGD, India · AIKosh national AI datasets platform

India's researchers and startups find verified datasets and models in one place, and train on them without renting GPUs.

IndiaAI wanted anyone building AI in India to get affordable, trustworthy and properly licensed data from government and private sources, and to experiment with it on the same platform.

India

What changed

1447datasets on one platform (as reported)
217AI models (as reported)
34contributing organisations (as reported)
What stood in the way

What stood in the way.

What stood in the way
  • Datasets were split across ministries, universities, hospitals and private firms, and engineers spent long hours finding, cleaning and validating them.
  • Private data was costly, quality and licensing varied, and GPUs for training were expensive to rent.
What changed
  • 1447datasets on one platform (as reported)
  • 217AI models (as reported)
  • 34contributing organisations (as reported)
What we built

What we built.

  • One searchable repository of datasets, models, toolkits and use cases, with Explorer, Contributor, Organisation Admin and Platform Admin roles.
  • A review by the organisation admin, a platform moderator and the Ministry before anything is published, and a clear licence on every dataset.
  • Ingestion by upload, push and pull APIs, SFTP and links to GitHub, Kaggle and Hugging Face.
  • Automated quality scores for uniqueness, completeness and updates, anonymisation tools and a GPU-enabled notebook for training and testing models.
  • GPU notebooks
  • Push and pull APIs
  • SFTP
  • GitHub, Kaggle and Hugging Face integrations
What it shows

What this work shows about us.

Trust built into the workflow

Three levels of review, a licence on every dataset and quality scores mean users know what they can rely on and how they may use it.

Built to grow with a country

More ministries and contributors can join through the same roles, approvals and ingestion paths.

Want to build a similar national AI datasets platform? Tell us what you want to change. Our AI assistant, Ask Unthinkable, helps you define your requirements, then connects you with the right team.

Start your plan ↗
If we built it today

If we built it today, with AI.

What we would add now, moment by moment.

Contributing data

An agent would check a new dataset before review, describe it, suggest the licence and point out missing fields or personal data to anonymise. The reviewers still approve every dataset.

Finding data

An agent would answer a researcher's question about what data exists for their problem, with the licence and quality score of each match. The researcher requests access through the normal approval.

Related

Where this work fits.

    Facts and figures from the project records; nothing added.

    Ask Unthinkable

    Want this for your business?

    Tell us what you run or what you're building. Our AI assistant, Ask Unthinkable, helps you define your requirements, then connects you with a senior AI engineer.

    Send us a note instead

    Required

    Thanks. A senior AI engineer will reply to you.
    u
    Ask UnthinkableClarity first, then the right team
    Your conversation

    An AI agent. This prototype uses prepared answers and sends nothing.