// bridge system

useState()→Model Weights·Event Propagation→Forward Pass·Array.map()→Tensor Operation·React diff→Loss Function L = Σ(y−ŷ)²·transition-duration→Learning Rate η·CSS clamp()→σ(x) Activation·Re-render cycle→Training Epoch·Event bubbling→Backpropagation ∂L/∂w·useCallback→Gradient Caching·Promise.all()→Batch Inference·Redux store→Weight Matrix·DevTools profiler→Loss Landscape·useState()→Model Weights·Event Propagation→Forward Pass·Array.map()→Tensor Operation·React diff→Loss Function L = Σ(y−ŷ)²·transition-duration→Learning Rate η·CSS clamp()→σ(x) Activation·Re-render cycle→Training Epoch·Event bubbling→Backpropagation ∂L/∂w·useCallback→Gradient Caching·Promise.all()→Batch Inference·Redux store→Weight Matrix·DevTools profiler→Loss Landscape·

GRID
Recommendation: hybrid architecture. Real-time anomaly detection on edge devices. Complex root-cause analysis in the cloud. Estimated efficiency gain: 73%. Bandwidth reduction: 94%.

Hassan Al-Rashid
GRID is right. Not everything needs to run on the edge, and not everything should run in the cloud. The architecture should match the requirements.

This is a decision you make every day as a frontend engineer. Client-side rendering or server-side rendering? The answer is always: it depends. Edge vs cloud inference follows the same decision framework.

Learning Objectives

○Evaluate when to use edge vs cloud inference
○Understand the tradeoffs: latency, model size, accuracy, connectivity, cost
○Design a hybrid edge-cloud architecture
○Map CSR/SSR decision frameworks to edge/cloud decisions

The Decision Matrix

Frontend

CSR vs SSR

// CSR: runs in browser, fast interaction
// SSR: runs on server, more capable

Machine Learning

Edge vs Cloud

// Edge: runs on device, low latency
// Cloud: runs on GPU, more capable

Structural Bridge

deployment-decision.tstypescript

type DeploymentTarget = 'edge' | 'cloud' | 'hybrid';

interface InferenceRequirements {
maxLatencyMs: number;
requiresOffline: boolean;
modelSizeMB: number;
accuracyThreshold: number;
dataPrivacy: 'public' | 'sensitive' | 'regulated';
}

function chooseDeployment(req: InferenceRequirements): DeploymentTarget {
// Must work offline → edge (like choosing CSR for offline PWA)
if (req.requiresOffline) return 'edge';

// Regulated data cannot leave the site → edge
if (req.dataPrivacy === 'regulated') return 'edge';

// Sub-100ms latency with unreliable network → edge
if (req.maxLatencyMs < 100) return 'edge';

// Model too large for device → cloud (like choosing SSR for heavy computation)
if (req.modelSizeMB > 300) return 'cloud';

// Real-time alerts on edge, deep analysis in cloud
return 'hybrid';
}

// Terra Grid examples:
console.log(chooseDeployment({
maxLatencyMs: 89,
requiresOffline: true,
modelSizeMB: 5,
accuracyThreshold: 0.95,
dataPrivacy: 'sensitive'
})); // 'edge' — vibration anomaly detection

console.log(chooseDeployment({
maxLatencyMs: 30000,
requiresOffline: false,
modelSizeMB: 2000,
accuracyThreshold: 0.99,
dataPrivacy: 'public'
})); // 'cloud' — root cause analysis

The hybrid pattern is the most common in production. Edge devices handle real-time detection and send compressed alerts to the cloud, where larger models perform deeper analysis. This mirrors the pattern of client-side validation with server-side verification.

Challenge

Design a deployment strategy for different Terra Grid inference tasks.

⚡

Exercise

Beginner~10 min

Design Compute Tradeoffs

Write a function chooseTarget(maxLatencyMs, requiresOffline, modelSizeMB) that returns 'edge' if offline is required OR latency < 100ms, returns 'cloud' if modelSizeMB > 300, and returns 'hybrid' otherwise.

function chooseTarget(maxLatencyMs, requiresOffline, modelSizeMB) {
  // Return 'edge', 'cloud', or 'hybrid'
  return null; // your code here
}

Key Takeaways

✓Edge vs cloud is a tradeoff, not a binary choice — like CSR vs SSR
✓Edge wins for: low latency, offline capability, data privacy, bandwidth reduction
✓Cloud wins for: complex models, high accuracy requirements, centralized analysis
✓Hybrid architectures combine edge detection with cloud analysis

Need a hint?

🧭 Guidance

✅ Solution

Edge vs Cloud Tradeoffs — Terra Grid | Tensorcraft

Edge vs Cloud Tradeoffs

Context