Présentiel et distanciel
3 days (21 hours)

Claude for Cloud and DevOps Engineers

AI-assisted Infrastructure as Code, troubleshooting of Kubernetes, OpenShift, Ceph and Proxmox, MCP for operations, CI automation and security guardrails (permissions, hooks). A course designed by architects who deploy private clouds in production, with labs on a dedicated environment and injected failures.

Training objectives

  • Install and secure Claude Code: permissions, hooks, organisation-managed settings.
  • Produce and review tested IaC with Claude (Ansible, Terraform, Helm, Kubernetes and OpenShift manifests).
  • Troubleshoot Kubernetes, OpenShift, Ceph, Proxmox and network incidents, read-only first.
  • Connect Claude to operations tools through MCP and develop an internal MCP server.
  • Automate PR reviews, reports, runbooks and post-mortems.
  • Protect an infrastructure agent against indirect injection and secret leakage.
  • Design a sovereign and compliant enterprise AI architecture.

Target audience

  • System, network and cloud engineers.
  • DevOps engineers and SREs.
  • Kubernetes, OpenShift, OpenStack, Proxmox and Ceph administrators.

Prerequisites

  • Linux administration and Git.
  • Basic knowledge of Kubernetes and Infrastructure as Code (Ansible or Terraform).
  • ECI-IA-101 recommended.

Detailed program

Day 1 — Claude Code in the infrastructure engineer's daily work

  • Refresher and ground rules: LLMs, context and hallucinations applied to operations, the roles of Claude Code, the Agent SDK and MCP, choice of model and effort, rule of engagement: the AI proposes, the engineer validates
  • Installing and securing Claude Code: installation and interfaces, permission modes, deny, ask and allow rules and their order of precedence, settings hierarchy from organisation to workstation, hooks, checkpoints and secrets kept out of the context. Labs 1 and 2: the infrastructure team's permission policy, blocking and logging hooks
  • CLAUDE.md and team standards: CLAUDE.md for an IaC repository (architecture, conventions, test commands), file hierarchy and imports, one session per topic. Lab 3: CLAUDE.md for the infra-lab repository
  • AI-assisted Infrastructure as Code: Ansible (roles, idempotency, molecule), Terraform and OpenTofu (reading a plan, drift), Kubernetes and OpenShift (Helm, Kustomize, SecurityContext, NetworkPolicy, SCC), pitfalls of obsolete APIs, invented options and dangerous defaults. Labs 4 to 6: Ansible hardening role, secured Helm application, analysis of a terraform plan

Day 2 — Troubleshooting, operations and MCP

  • AI-assisted diagnosis and troubleshooting: method from symptom to fix, Kubernetes and OpenShift (CrashLoopBackOff, Pending, DNS, degraded operators), Ceph (HEALTH_WARN, degraded PGs, OSD down, near-full), Proxmox (Corosync quorum, HA), network (MTU, VLAN, bonding, VXLAN), filtering large volumes of logs. Labs 7 to 9: three Kubernetes failures, Ceph HEALTH_WARN, loss of Proxmox quorum
  • MCP for operations: read-only MCP servers for Kubernetes, Git, Prometheus and ticketing, stdio and Streamable HTTP transports, security (local listening, origin validation, authentication), configuration scopes and a versioned .mcp.json file. Labs 10 and 11: Kubernetes and Prometheus MCP, a Ceph ops MCP server
  • Runbooks, post-mortems and documentation: runbooks generated from real procedures, blameless post-mortem from the logs, architecture diagrams and flow matrices, team skills. Lab 12: post-mortem skill applied to the Ceph incident

Day 3 — Automation, architecture and governance

  • Claude Code in CI/CD and operations agents: non-interactive mode and JSON output, GitHub Actions and GitLab CI, automatic review of infrastructure PRs, scheduled agents (health report, capacity analysis), human validation before any write. Labs 13 and 14: Terraform and Helm PR review in CI, daily report agent
  • Agent security, architecture and sovereignty: indirect injection through logs, tickets and web pages, least privilege, isolation and human validation, secrets management, ways of accessing models, open models on premises, VRAM, quantisation and GPUs, Law 09-08 and CNDP, Law 2013-450 and ARTCI, sector requirements. Workshop: AI architecture for a telecom operator or a bank
  • Final scenario and governance: operations scenarios, usage policy for the infrastructure team's AI agents. Graded 60-minute incident scenario, under strict permissions

Format and assessment

  • On site or live online, 75% hands-on on a dedicated lab; in-house sessions on request
  • Delivered in French
  • 14 labs, 10 scenarios and 1 graded final incident scenario
  • Assessment: quiz (10%), validated labs (35%), scenarios (10%), architecture workshop (10%), final incident scenario (35%)
  • ECINTELLIGENCE training attestation awarded from a score of 60%

Materials provided to participants

  • ECINTELLIGENCE course manual and lab workbook (PDF)
  • Templates: permissions file, CLAUDE.md for an IaC repository, security hooks, post-mortem skill, CI pipeline
  • Code of the Ceph ops MCP server
  • Cheat sheet: AI-assisted diagnosis (Kubernetes, OpenShift, Ceph, Proxmox)
  • AI agent usage policy template
  • AI glossary sheet and VRAM sizing table
  • Dated state-of-the-art appendix

Next steps

As models, versions and features evolve quickly, the content is updated before each session. Product names mentioned (PyTorch, OpenShift, Kubernetes, Proxmox, NVIDIA, Hugging Face…) are trademarks of their respective owners.

Independent training designed by ECINTELLIGENCE. Claude and Anthropic are trademarks of Anthropic. This training is neither delivered nor endorsed by Anthropic and does not prepare for any official Anthropic certification.

Certification

At the end of this training, you will receive a certificate of participation issued by squint.

Price on request

Duration

3 days (21 hours)

Format

Présentiel et distanciel

Next session

On request

Request a quote

Other training courses that might interest you

AI Landscape for Decision-Makers

Présentiel et distanciel

A jargon-free day for decision-makers: an overview of predictive, generative and agentic AI, what makes a project succeed, how to …

1 day (7 hours) Learn more

Claude for Professionals

Présentiel et distanciel

A hands-on course for every profession: writing effective requests, choosing the right model, making the most of projects, documents, search …

2 days (14 hours) Learn more

Building with the Claude API and Agents

Présentiel et distanciel

A developer course: Messages API, streaming, errors and limits, cost control (batch, prompt caching), structured outputs, tools, MCP, Agent SDK, …

4 days (28 hours) Learn more

Ready to develop your skills?

Join hundreds of professionals who have trusted squint for their skills.

View all our training courses