AI Integration Tutorial

1. Setting Up Your AI Project

Initialize Project

Create a new directory for your AI model and install the required dependencies using our decentralized SDK.


mkdir my-ai-model
cd my-ai-model
npm init -y
npm install @ezeniiaadeploy/sdk

Create Configuration

Configure your decentralized training settings with node selection and security parameters.


// config.json
{
  "training": {
    "epochs": 100,
    "compute_nodes": ["node-01", "node-02"],
    "security": {
      "signature_key": "0x3E..."
    }
  },
  "model": "ezeniia-ml-v2"
}

2. Implementing Your Training Script

Create Training Script

Use the Ezeni Iia SDK to distribute your training process across multiple nodes for faster execution.


import { TrainController } from '@ezeniia/sdks';

const controller = new TrainController({
  config: 'config.json',
  dataset: 's3://ezeniia/datasets/image-net',
  modelType: 'vision-transformer'
});

await controller.start(); // Initializes parallel training

Training Progress:
[node-01] ✓ Layer 1: 32% (6/18)
[node-02] ▶ Layer 2: 18% (4/22)
Total Sync Time: 02:35:17
Remaining: ~2 hrs 42 mins

Continue to Next Step →

3. Monitoring and Optimization

Training Dashboard

Access real-time metrics showing GPU utilization and validation accuracy across all nodes.

View in Dashboard

Parameter Tuning

Adjust hyperparameters like learning rate and batch size based on distributed performance metrics.

+12% accuracy since last optimization

Node Performance

• node-01: 92% efficient
• node-02: ! Needs rebalancing
• node-03: ✅ Optimal sync

View Performance Metrics ↓

4. Deployment and Iteration

Model Deployment

When your model reaches optimal performance, use the framework to deploy it across the decentralized network.


const deployer = new ModelDeployer({
  modelPath: './trained_model',
  nodes: ['node-01', 'node-03']
});

await deployer.push(); // Distributes model across 3+ nodes

Continuous Training

Configure auto-refresh settings to retrain your model as new data becomes available in the network.

Auto-deploy enabled: ON