Architecture
The Darwin platform is a state-of-the-art AI layer blockchain that integrates distributed AI inference capabilities, fostering a robust, scalable, and secure AI development and deployment environment. Built on an EVM-compatible blockchain, it allows seamless integration with Ethereum-based tools and infrastructures.
Key Components of the Darwin System
Darwin's main architecture contains three different layers:
Data Layer
The data layer is divided into two storage types:
Temporary Storage: It is designed for high performance and utilized during training or inferencing phases.
Permanent Storage: This is a more cost-effective solution for long-term data storage. Third-party services are employed for permanent storage to optimize costs.
System Layer
The main layer comprises of three key components: Blockchain, Query Mixer, and AI Node.
a. Blockchain
Stake Pool and Quota System: Users stake $DNA tokens to gain access to AI inference quotas, which helps secure the network and enables decentralized governance.
Transaction Recording: All transactions, whether AI-related or not, are permanently recorded on the blockchain.
Darwin Contracts: These smart contracts facilitate the submission and validation of AI transactions, ensuring secure and compliant operations.
Gasless Transactions: The system provides a gasless AI service using the "Approve-Once-Approaches-on-Call" method, delivering a smoother user experience.
b. Query Mixer Network
Optimization and Anonymization: This component optimizes and anonymizes AI task processing, ensuring transparent and unbiased operations across distributed AI nodes.
Resource Allocation: It efficiently allocates AI resources throughout the network, balancing the load and optimizing performance.
c. Distributed AI Inference
AI Nodes: These form the backbone of the AI inference system, efficiently handling complex AI tasks.
SPECTER: Darwin's zk-proof engine enhances privacy and security by generating zero-knowledge proofs for AI inferences.
Application Layer
The application layer includes the following components:
Token/Transaction Operations: This covers operations related to tokens or transactions, including interactions through wallets and blockchain explorers.
DApps: Decentralized applications like DeFi and AI-powered services.
AI Node Client: Users can participate in the AI system through this client.
System Roles and Interaction Flow
The Darwin AI blockchain platform architecture integrates users, relayers, chain nodes, query mixers, and AI nodes into a cohesive system. These interactions are organized into three operational circles:
Note:
Each circle plays a vital role in the platform's AI task execution.
Frontend on Chain
Users and Wallet Interaction:
Users interact with the blockchain via EVM-compatible wallets.
They stake $DNA tokens for inference quotas to submit AI transactions without direct cost.
Transactions can be submitted directly or through relayers to reduce gas fees.
Relayers:
Relayers is an API service that implements gasless functionality. It allows users to send AI transactions for free by paying with $DNA tokens.
Relayers bundle multiple user requests and forward them to chain nodes, facilitating gasless transactions and reducing user costs.
Backend on Chain
Chain Nodes:
Process transactions from users or relayers.
Handle transaction verification, smart contract execution, and maintain blockchain security.
Staking and Resource Allocation:
Manage staking and resource allocation to ensure network security and resource distribution.
Wallet addresses linked to chain nodes track stakes and resource allocation.
Functioned AI
Query Mixer Operations:
Monitors on-chain events to process AI transactions securely.
Generates additional queries to validate AI transaction outputs.
AI Nodes and Real-time Inference:
AI nodes process tasks from the Query Mixer to generate real-time AI inferences.
Results are streamed back to users for a responsive AI service.
Incentives and Verification:
AI nodes and the Query Mixer stake $DNA tokens to ensure commitment and reliability.
Verifiable Random Functions (VRF) perform random checks on AI nodes for model accuracy and latency, enforcing high operational standards.
Interaction Dynamics
Staked quotas from users are allocated to AI nodes by the Query Mixer as rewards.
All participants must register and stake $DNA tokens to participate and receive tasks.
Continuous monitoring and penalties ensure adherence to operational standards, maintaining a robust AI processing environment.
This structure ensures a secure, efficient, and user-friendly AI blockchain platform, supporting scalable and economically viable AI and blockchain integration.
Last updated
