Skip to content
logo
Sensor AI
  • Pages
    • Introduction
    • Core Features
      • Annotation Tool
      • Crowdsourced Annotation
      • Dataset Marketplace
      • AI Model Training
    • Roadmap
    • Tokenomics
      • Token Utilities
      • Token Distribution
      • Governance
    • icon picker
      Technical Documentation
    • Use Cases
    • Tutorials
      • Annotating a dataset
      • Selling a dataset
      • Training a custom model
    • FAQ
    • Links
    • Resources
      • Privacy Policy
      • Terms of Service
      • Disclaimer
logo
Sensor AI
/
Technical Documentation

Technical Documentation

Sensor AI’s architecture is designed to ensure decentralization, security, and scalability while maintaining usability. The system consists of three primary layers:

1. Blockchain Layer (Ethereum)

Role: Manages transactions, governance, and platform logic via smart contracts.
Key Components:
$SENSE Token (ERC-20): Used for payments, staking, and governance.
Smart Contracts: Handle annotation rewards, dataset marketplace transactions, and voting mechanisms.
Gas Optimization: Implements batch transactions and layer-2 solutions (e.g., Polygon) to reduce costs.

2. Storage Layer (IPFS & Filecoin)

Role: Stores datasets and annotations off-chain while ensuring immutability.
Key Components:
IPFS (InterPlanetary File System): Decentralized storage for raw images and labeled datasets.
Filecoin Integration: Optional long-term, incentivized storage.
On-Chain References: Each dataset’s IPFS hash is recorded on Ethereum for verification.

3. Compute Layer (Decentralized AI Training)

Role: Provides distributed computational power for model training.
Key Components:
Golem/iExec Integration: Leverages decentralized GPU/CPU networks for cost-efficient training.
Task Coordination: Smart contracts assign jobs and release $SENSE payments upon completion.
Privacy: Data remains encrypted during processing (secure enclaves where possible).
Want to print your doc?
This is not the way.
Try clicking the ··· in the right corner or using a keyboard shortcut (
CtrlP
) instead.