Inspiration

We started KoData DAO with a simple but urgent question:

“Who owns the data that trains the world’s AI models?”

In 2030, an AI model trained on biased or incomplete data could decide who gets a loan, a job, or life-saving healthcare. Yet, most of the world’s datasets today are labeled far from the contexts they’re meant to serve; leaving African, Asian, and Latin American environments digitally invisible.

We built KoData DAO to solve this: a decentralized data-labeling network that empowers local people to earn crypto by labeling local data; from forests and crops to dialects and traffic scenes; while ensuring verifiable provenance and fair compensation through Starknet. While using xVerse as the token's wallet system.

Our belief is simple: AI should be locally trained, globally trusted.

What it does

KoData DAO is a cross-chain decentralized data labeling and governance system, built to make data ownership transparent, provable, and participatory.

It has two key layers:

  1. Data Labeling & Curation Layer (Starknet + IPFS)
  2. Contributors label text, image, or sensor data through our task interface.
  3. Submissions are verified via decentralized queues and stored on IPFS for permanence.
  4. Validated contributors earn MAD (Machine/AI Data) tokens as proof of contribution and reward.

  5. Governance & Ownership Layer (Bitcoin Runes + Starknet DAO)

  6. Each dataset becomes a governed asset.

  7. Contributors and Runes holders can create proposals, vote, and manage dataset usage, gasslessly, via Starknet’s Account Abstraction.

  8. Governance actions are recorded on-chain and represented through Runes-based DAO tokens, enabling verifiable voting and transparent treasury management.

Together, these layers create a living, self-governing Data DAO; one that rewards contributors, enforces provenance, and gives communities a say in how data fuels AI.

KoData DAO is a decentralized autonomous organization that manages community-driven data labeling tasks on Starknet.

Essentially, KoData DAO is “Proof of Label” for the AI era; a trustless pipeline for clean, contextual, and community-owned data.

How we built it

  • Smart Contracts: Written in Cairo, deployed on Starknet to handle staking, payouts, and DAO voting.

  • Off-chain Orchestration: Implemented with Node.js (Express) and Redis/BullMQ to manage labeling task queues and worker validation jobs.

  • Wallet & Verification: Integrated xVerse for Starknet wallet support, token storage, and participant verification.

  • ZK validation: Integrated zero-knowledge circuits to confirm labeling integrity (using validity proofs rather than rechecking all data).

  • Frontend: Built with Next.js and Starknet.js for wallet connection and DAO participation.

  • Data storage: Tasks and proofs are stored via IPFS, while only hashes and reward logic live on-chain.

Challenges we ran into

  • Integrating Runes metadata with Starknet governance logic required custom state sync mechanisms.

  • Designing fair reward models that balance gas efficiency with contributor volume.

  • Integrating ZK proofs for off-chain work without compromising performance.

  • Managing quality assurance among distributed human labelers.

  • Making the UX friendly enough for non-crypto-native African participants.

  • Learning Cairo and the starknet toolchain efficiently.

Accomplishments that we're proud of

  • Deployed the first on-chain data-labeling DAO prototype on Starknet.

  • Live prototype connecting Xverse Wallet with Starknet DAO contracts.

  • Validated with real local labelers who completed environmental image annotations.

  • Created a Proof-of-Label token model, linking task validation directly to Starknet smart contracts.

  • Built a sustainable earn-and-train loop; where labelers become dataset validators and later DAO voters.

What we learned

  • Data sovereignty isn’t just a technical issue; it’s a governance, cultural, and economic one.

  • Decentralization works best when aligned with human incentives; not just tokenomics.

  • Starknet’s ZK scalability and Cairo’s expressiveness make it ideal for privacy-preserving human verification.

  • The next AI revolution won’t come from bigger models, but from fairer, context-rich data.

What's next for KoData DAO

  • Complete the launch of our DAO governance with labeler representation.

  • Integrate on-chain reputation scoring for consistent contributors.

  • Partner with AI labs and NGOs working on climate, agriculture, and healthcare data.

  • Build KoData Grants, funding local data projects across Africa using DAO treasury votes.

  • Open-source our Proof-of-Label Cairo module for other Starknet developers.

Built With

Share this project:

Updates