Skip to main content
🎤 Luca Berton is speaking at Red Hat Summit & KubeCon EU 2026!Learn more →
  • Instant download
  • Lifetime updates
  • Secure Stripe checkout
  • EU VAT invoice
Ebook
CopyPasteLearn

RHEL AI Engineering Recipes

Production cookbook for Red Hat AI — the recipes platform and SRE engineers need to ship InstructLab fine-tunes, vLLM serving, RAG, and multi-agent stacks on RHEL 9 GPU nodes. What's inside: • Subscr

Loved by DevOps practitioners
One-time purchase
€49.00EUR
Tax shown at checkout · No subscription
  • Production-tested examples you can copy-paste today
  • PDF format, optimized for screen and print
  • Free updates whenever the source CLI changes
  • Search-friendly, code-block heavy, no fluff

About this ebook

Production cookbook for Red Hat AI — the recipes platform and SRE engineers need to ship InstructLab fine-tunes, vLLM serving, RAG, and multi-agent stacks on RHEL 9 GPU nodes.

What's inside:

• Subscription-manager + repo enablement for `ilab`, CUDA, and Intel Gaudi accelerators.

• GPU smoke tests and troubleshooting recipes (nvidia-smi, CDI device resolution, persistent mode, ECC).

• Cloud bootstrap recipes for AWS p5/g5, Azure ND H100 v5, GCP A3, and IBM Cloud GX3.

• Bare-metal kickstart for PXE-provisioned RHEL AI nodes.

• InstructLab taxonomy templates (knowledge + skills), SDG runbook, multi-GPU DeepSpeed ZeRO-3 training.

• vLLM serving with tensor parallel, chunked prefill, GPU memory tuning, and OpenAI-compatible inference.

• FastAPI wrapper, ChromaDB + LangChain RAG pipeline, CrewAI multi-agent template.

• Idempotent Ansible role to deploy RHEL AI across a fleet, vaulted activation keys included.

• Observability: nvidia-dcgm-exporter, vLLM /metrics scrape config, SLO Prometheus rules, Grafana alerts.

• Scale-out topologies: data-parallel vs tensor-parallel vs pipeline-parallel, multi-node DeepSpeed + Ray.

• Governance: SPDX lineage YAML, model cards, cosign keyless artifact signing, request-level audit logs.

• Blue-green and canary model rollouts via weighted upstreams.

• Hardware sizing reference (VRAM per parameter, tokens/sec per GPU class) and 10 common-error fixes.

Tested against RHEL AI 1.5 on A100, H100, and L40S GPUs. Lifetime updates while the recipes are maintained, delivered via /library.

Before you buy

  • Instant digital delivery — by completing checkout you consent to immediate access. See the digital delivery policy.
  • Refunds available within 14 days if you have not yet downloaded the file. Full terms in the refund policy.
  • Your purchase email becomes your library sign-in. Tax is calculated and shown on the next step.