Agentic AI Reference Architecture

Model Adaptation · Software component

Federated Aggregator

Software componentModel AdaptationModelsarc:FederatedAggregator

A central training component that distributes the global model to devices and aggregates their locally computed updates, optionally with differential-privacy noise, into an improved global model.

Responsibility. Improves a global model from distributed device updates without collecting raw data.

Also known as: Federated learning server

orchestrates; receives data fromtrainsOn-Device Trainer: orchestrates; receives data fromOn-Device TrainerEdge-Optimized Model: trainsEdge-Optimized Model
Direct neighbourhood (hover for relationship types)

Relationships

receives data from dynamic

orchestrates control

trains lifecycle

Classification

Patterns
Federated averagingDifferential privacy
Quality attributes
Privacy (NIST AI RMF: privacy-enhanced)
Risks mitigated
Re-identification of individual device data

Sources

  1. Ch4.3: T. Nguyen, "Container Orchestration and Edge Deployment," in Mastering Agentic AI Systems: Guide for the NVIDIA NCP-AAI Exam, 1st ed. 2026, ch. 4.3. ISBN: 9798244538229.