Observe
AI analyzes and explains. No changes are made. Useful for understanding a project.
AI-Assisted DevOps Mastery // V1 Skeleton
Learn DevOps by doing the work.
Created by William Elias, DevOps Learn by Doing is a CLI learning and engineering platform built around a real production-style Python API project. The AI works with you, explains what matters, and never pretends to know what it has not actually done. Progress means you can direct, review, troubleshoot, and improve AI-assisted infrastructure—not type every command from memory.
C:\LEARN> devops-learn local projects/api_platform
[SYSTEM] Production-Style API Platform
[LANG] Python
[DOCKER] (real)
[CLOUD] Azure (simulated)
[MODE] Collaborative + learning depth
[TRACK] Inspect → Tests → Lint → Docker → Verify
→ Terraform → Azure → Kubernetes
→ Failure → Troubleshoot → Rollback
[GOAL] Understand enough to direct and review the AI
[READY] No cloud credentials required for local work.
C:\LEARN>
01 // Learning model
AI-assisted DevOps mastery.
The AI and the human do real engineering work together. The platform explains concepts in context, targets gaps in the learner profile, and treats competency as the ability to direct, review, troubleshoot, and improve AI-generated work.
Guided
AI explains the next step; the human performs meaningful actions and learns by doing.
Collaborative
AI generates and performs substantial work while involving the human in decisions.
AI-Executed / Autonomous
AI performs approved work and narrates what, why, risks, and rollback. Destructive or costly actions still require human approval.
Introduce→
Explain→
Predict→
Perform→
Validate→
Explain result→
Challenge
02 // Training path
One project. A full DevOps journey.
V1 follows a production-style API platform from local development through delivery, infrastructure concepts, operations, failure recovery, and architecture review.
FastAPI→
Git→
Tests→
Docker→
GitHub Actions→
Terraform→
Azure networking→
Kubernetes→
Identity + secrets→
Observability→
Deploy→
Break it→
Troubleshoot→
Rollback→
Architecture review
03 // Truthful status
Real local execution. Real-only Azure lifecycle.
The boundary is deliberate: local work, saved Terraform plans, and the Azure lifecycle implementation are real-only. Azure execution remains opt-in and is never claimed as verified until a credentialed run succeeds.
Real
Project analysis, learner profile, SQLite persistence, audit, and experience tracking.
Real
Local vertical slice: pytest, flake8, docker build/run/logs/stop, and HTTP health checks.
Real
The demo FastAPI application and its tests are runnable code.
Sim
Azure provisioning and Terraform apply require explicit approval and opt-in credentials; Kubernetes remains simulated.
Controlled execution
The AI provider cannot invoke tools directly. ToolService.invoke is the single controlled entry point,
with risk metadata and approval gating built into the tool model.
Architecture
A modular monolith, not an agent swarm: explicit service boundaries, constructor injection, plain domain models, and SQLite persistence with SQL isolated behind repositories.
Extension points
Azure and Python are the implemented V1 path. AWS, GCP, and Go have typed extension points, but their curricula are intentionally not claimed as complete yet.
04 // Roadmap
Move from simulation to real execution in controlled steps.
The next milestones keep the existing safety boundary and replace simulated tools one layer at a time.
Real Python + Docker
Done Run pytest, flake8, and the local Docker daemon through narrow allow-listed subprocess wrappers.
Real Terraform validation
Use saved Terraform plans with explicit approval before real-only apply and verified cleanup.
Azure lifecycle acceptance run
Exercise the real, approval-controlled Container Apps path and independently verify cleanup.
05 // Technical Q&A
AI-assisted platform engineering & safe automation.
Key architectural questions on operating infrastructure, security gates, and autonomy boundaries alongside AI agents.
How does AI-assisted DevOps training bridge the gap to production engineering?
Traditional courses rely on contrived browser sandboxes that abstract away real tooling friction. DevOps Learn by Doing grounds learners in a real production-grade Python API codebase, real Docker build contexts, and real Terraform manifests. Instead of memorizing commands, engineers learn how to direct, verify, troubleshoot, and safely govern AI-assisted infrastructure operations.
Why use security gates and sandbox modes for infrastructure automation?
Unconstrained agent execution against cloud providers risks runaway resource provisioning, accidental data exposure, or credential leakage. DevOps Learn by Doing enforces security-gated verification using Trivy vulnerability scanning, Conftest policy-as-code assertions, and read-only dry-run simulation modes before any state-mutating deployment can execute.
How does the four-level autonomy model prevent dangerous AI mutations?
By categorizing operational tasks into Observe, Guided, Collaborative, and Autonomous tiers, the system ensures that irreversible actions (such as Terraform applies, database migrations, or remote cloud destruction) require cryptographic human sign-off. AI agents assist with drafting and analysis, while humans retain authority over production state.
What core technical competencies are required for platform engineers working with AI?
As generative models increasingly write deployment scripts and configuration files, the vital platform engineering skill shifts from syntax memorization to rigorous validation: evaluating diff blast radius, writing declarative policies, inspecting container isolation, interpreting structured audit logs, and defining deterministic rollback protocols.
# 01 CHECK
devops-learn doctor
# 02 UNDERSTAND
devops-learn init .
# 03 BUILD
devops-learn local .
# 04 SECURE
devops-learn security scan .
# 05 DEPLOY
devops-learn deploy .
# 06 PROVE
devops-learn report