[LFX Mentorship] Expand HAMi GPU-sharing tutorials, labs, and troubleshooting guides

[rootsongjc]

Summary

Create and improve practical, reproducible learning materials that help users deploy, validate, and troubleshoot HAMi GPU sharing in realistic Kubernetes and AI workloads.

This issue is the upstream tracking issue for a proposed CNCF LFX Mentorship program. It builds on the tutorials/labs work tracked in #431 and should contribute to the maintained HAMi website and documentation. The former Project-HAMi/hami-workshop repository is archived, so new work should land in active HAMi repositories.

Scope

• Audit the current HAMi tutorials, demos, labs, and troubleshooting content and identify the highest-impact gaps.
• Add reproducible learning paths for installing HAMi and validating GPU sharing and isolation.
• Add practical examples for representative AI workloads such as vLLM, Ray, or SGLang, based on mentor-approved priorities.
• Document common failure modes, diagnostic commands, expected results, and recovery steps.
• Improve navigation and cross-links so users can move from concepts to hands-on labs.
• Validate examples against a documented environment and supported HAMi release.

Expected deliverables

• A short gap analysis and mentor-approved implementation plan.
• At least two substantial, reproducible labs or tutorials.
• A troubleshooting guide covering common installation, scheduling, and runtime issues.
• Tested manifests/scripts and clearly documented prerequisites.
• Contributions merged into the active HAMi website or another mentor-approved HAMi repository.

Acceptance criteria

• Instructions are reproducible from a clean environment with the documented prerequisites.
• Each lab includes learning goals, expected output, verification steps, and cleanup instructions.
• Commands and manifests pass the relevant repository checks.
• Content is reviewed for technical accuracy and discoverability by HAMi maintainers.
• Final documentation states the HAMi version and hardware/software assumptions used for validation.

Out of scope

• Reviving the archived Project-HAMi/hami-workshop repository.
• Building a new documentation platform.
• Promising support for every accelerator, framework, or deployment model within one mentorship term.

Related work

• Proposal: Add Tutorials, Demos & Labs Section #431
• https://github.com/Project-HAMi/hami-workshop

[coderabbitai]

🔗 Related PRs

#530 - docs: update CNCF status from Sandbox to Incubating [merged]
#560 - Add Lab 7: GPU isolation on k3s without the GPU Operator [merged]
#608 - docs: add Kueue and HAMi vGPU quota lab [merged]
#610 - docs: add lab 10 topology-aware scheduling tutorial [merged]
#654 - docs(tutorials): add Lab 11 — Run SGLang on HAMi GPU Shares [open]

---

📝 Issue Planner

_{Check the box below or use the @coderabbitai plan command to generate an implementation plan and prompts that you can use with your favorite coding assistant.}

• Create Plan

---

🧪 Issue enrichment is currently in open beta.

You can configure auto-planning by selecting labels in the issue_enrichment configuration.

To disable automatic issue enrichment, add the following to your .coderabbit.yaml:

issue_enrichment:
  auto_enrich:
    enabled: false

💬 Have feedback or questions? Drop into our discord!