Skip to main content

Part 7: AI Cloud Native Development

You've built AI agents locally in Parts 1-6—designing agent architectures, implementing integrations, and mastering spec-driven development. Part 7 bridges local development to production deployment at scale. You'll learn containerization strategies, orchestration patterns, and operational excellence practices that transform agents from learning projects into production systems serving real users.

Prerequisites: Parts 4-5 (Python fundamentals and spec-driven development) are required. Part 6 (AI Native Software Development) is strongly recommended for agent building experience before tackling deployment.

Goals​

By completing Part 7, you will:

  • Understand agent deployment architectures: Learn how containerized agents, orchestration platforms (Kubernetes, Dapr), and infrastructure patterns enable production-scale systems
  • Implement containerization strategies: Package agents with Docker, manage multi-stage builds, and optimize images for deployment efficiency
  • Apply orchestration patterns: Deploy agents to Kubernetes clusters, manage state with Dapr Actors, and orchestrate workflows with Dapr's distributed application runtime
  • Design operational excellence: Build observability into agent systems (metrics, logs, traces), engineer for cost optimization, and implement security governance for production deployments

Chapter Progression​

Part 7's 12 chapters build deployment capability through four stages:

Foundational Infrastructure (Chapters 50-52)​

Start with production API patterns and containerization fundamentals.

  • Chapter 50: FastAPI for AI Cloud-Native Services (Deep Dive) — Advanced API patterns, async processing, dependency injection, and deployment configurations
  • Chapter 51: Docker for AI Services — Building container images, multi-stage builds, optimization strategies, and shipping agents as containers
  • Chapter 52: Apache Kafka for Event-Driven AI Systems — Event streaming for agent communication, message patterns, and asynchronous architectures

Orchestration (Chapters 53-54)​

Scale from single containers to orchestrated clusters with deployment automation.

  • Chapter 53: Kubernetes for AI Services — Container orchestration, deployments, services, config management, and running agents in K8s clusters
  • Chapter 54: CI/CD & Infrastructure-as-Code for AI Services — Automated pipelines, infrastructure provisioning, GitOps patterns, and deployment workflows

Dapr Framework (Chapters 55-58)​

Apply Dapr's distributed application runtime for microservices, state, workflows, and agent-specific patterns.

  • Chapter 55: Dapr for AI Microservices — Sidecar building blocks, service invocation, pub/sub messaging, and state management
  • Chapter 56: Dapr Actors for Agentic State and Concurrency — Virtual actor pattern for agent state, concurrency control, and distributed coordination
  • Chapter 57: Dapr Workflows for Long-Running Orchestration — Durable workflows, saga patterns, and long-running agent processes
  • Chapter 58: Dapr Agents — Designing agentic services on Dapr, combining actors + workflows + AI for production agent architectures

Operations Excellence (Chapters 59-61)​

Operate agent systems with observability, security, and cost engineering.

  • Chapter 59: Observability, Cost & Performance Engineering — Metrics, logs, traces, cost optimization strategies, and performance tuning for agent workloads
  • Chapter 60: API Edge & Gateway for AI Services — Ingress patterns, Kong gateway configuration, rate limiting, and API management
  • Chapter 61: Security, Safety & Governance for Agentic Systems — Authentication, authorization, secrets management, safety guardrails, and compliance

Why this sequence? You can't orchestrate what you can't containerize. You can't operate what you haven't deployed. Each stage builds deployment capability: Foundation (can containerize agents) → Orchestration (can deploy to clusters) → Dapr Framework (can build distributed agent systems) → Operations (can run production systems reliably and securely).