Skip to content

Tier 3: Operations and Scale — About This Tier

You’ve shipped code with AI assistance (Tier 1) and built complex systems (Tier 2). Now you own the production systems themselves. Tier 3 is for DevOps engineers, tech leads, and senior developers who ask: “What happens after code ships? How do we monitor it? How do we scale this across our team? How do we stay ahead of tool evolution?”

This tier shifts focus from building to operating—and from individual developer workflows to organization-wide adoption. You’ll learn how AI changes incident response, automate CI/CD pipelines with Claude Code, establish team conventions that prevent chaos, and build an evaluation framework for whatever comes next in the rapidly evolving AI tooling landscape.

Tier 3 answers two fundamental questions:

  1. How do we keep AI-generated systems reliable in production?
  2. How do we scale AI-augmented workflows across teams without loss of control?
ModuleFocusDuration
M11: Post-DeploymentMonitoring, observability, incident response~90 min
M12: CI/CD IntegrationPipeline automation, headless workflows, batching~90 min
M13: Team AdoptionStandards, safety, scaling to teams~90 min
M14: What’s NextTool evaluation, future of roles, transferable principles~90 min

Total Time Commitment: ~5.5 hours (spread across multiple sessions)

Each module follows this rhythm:

  • Pre-work (15-20 min): Self-paced reading on theory and fundamentals
  • Self-directed Workshop (45-60 min): Hands-on exercises you work through at your own pace
  • Concrete Takeaway: An artifact you’ll actually use (workflow, playbook, config, framework)

You must have completed Tiers 1 and 2 before starting Tier 3. Specifically:

  • Comfortable using Claude Code and basic prompt engineering
  • Familiar with agent patterns and MCP integrations
  • Experience deploying and running systems
  • Understand Git workflows and deployment processes

By the end of Tier 3, you will:

  • Diagnose production issues faster using AI-augmented incident investigation
  • Automate CI/CD pipelines with Claude Code in headless mode
  • Establish team conventions and permission policies for safe, scaled AI workflows
  • Evaluate new AI tools using a principled framework that survives tool changes
  • Answer “what’s next?” with confidence rather than chasing hype

From Individual to Organizational: The leap from “I use Claude Code” to “our team uses Claude Code” requires safety guardrails, cost controls, and clear conventions. We’ll cover all three.

The Principle: “Developer Owns All Code”: AI can generate code faster, but developers remain accountable for correctness, security, and maintainability. This principle shapes team policy throughout Tier 3.

Observability Over Hope: You can’t manage what you can’t measure. This applies to system reliability, team adoption, costs, and tool effectiveness.

Transferable Principles: The specific tools will change. The principles—context management, agent design, tool composition, security—persist. Tier 3 teaches you to think about tools, not just use them.

  • Solo path: Work through modules in order; complete pre-work before workshop sessions
  • Team path: Run sessions synchronously; discuss as a group during workshops
  • Asynchronous path: Teams self-organize exercises and share results
  • M11 resonates most with on-call engineers and SREs; consider inviting them as guest perspectives
  • M12 appeals to platform/infra teams; it’s the most immediately deployable module
  • M13 requires organizational context; adapt exercises to your actual team structure and tools
  • M14 is philosophical but grounded; use it as capstone synthesis, not afterthought

Next Steps: Start with M11: Post-Deployment