AI-Powered Developer Platform on ROSA

Welcome to the AI-Powered Developer Platform on ROSA workshop. In this half-day hands-on lab, you will learn how Red Hat OpenShift Service on AWS (ROSA) provides the perfect foundation for modern, AI-assisted software delivery — from running GPU-backed Large Language Models as a managed service, to giving every developer an intelligent cloud IDE that generates, deploys, and audits their applications automatically.

Who This Workshop Is For

This workshop is designed for two complementary personas that will work through the labs together:

Persona	Role in This Workshop
Platform Engineer	Deploys ROSA infrastructure, provisions scalable GPU capacity, deploys an LLM as a Service using OpenShift AI, and sets the developer platform conventions — all without writing a runbook.
Developer	Launches a cloud IDE (OpenShift Dev Spaces), uses an AI coding assistant (OpenCode) backed by the in-cluster LLM, generates and deploys a real application, then validates it against platform rules — all without leaving the browser.

Persona

Role in This Workshop

Platform Engineer

Deploys ROSA infrastructure, provisions scalable GPU capacity, deploys an LLM as a Service using OpenShift AI, and sets the developer platform conventions — all without writing a runbook.

Developer

Launches a cloud IDE (OpenShift Dev Spaces), uses an AI coding assistant (OpenCode) backed by the in-cluster LLM, generates and deploys a real application, then validates it against platform rules — all without leaving the browser.

What You Will Build

By the end of this workshop, you will have deployed a complete AI developer platform on ROSA and shipped a real application through it:

Platform Engineer Track

A fixed-size GPU machine pool on ROSA — one GPU node, no manual EC2 interaction
OpenShift AI (RHOAI) with KServe serving Qwen3.6-35B (a state-of-the-art MoE reasoning model) via a fully OpenAI-compatible REST API
OpenShift Dev Spaces — a browser-based cloud IDE pre-configured for every developer
AGENTS.md — a machine-readable rulebook that makes OpenCode enforce your platform conventions automatically
A Go application template — the "golden template" developers clone to start new projects
OpenShift GitOps (Argo CD) and OpenShift Pipelines (Tekton) — GitOps delivery infrastructure, ready for teams

Developer Track

A personal Git repository — provisioned in seconds via the Git server, no GitHub account needed
A complete Fortune Cookie application generated live by OpenCode + Qwen3.6: Go binary, web UI, Kustomize manifests
The app built and deployed automatically via OpenShift Pipelines and Argo CD
Platform compliance validated — and auto-fixed — by the same LLM that wrote the code

Learning Outcomes

After completing this workshop, you will understand:

How ROSA’s managed machine pools give you on-demand GPU capacity without EC2 complexity
How OpenShift AI (KServe + vLLM) provides a scalable, OpenAI-compatible LLM API on your own cluster
How AGENTS.md shifts platform governance from documentation to code
How OpenShift Dev Spaces + OpenCode creates an AI-augmented developer experience that requires no local setup
How OpenShift Pipelines and Argo CD form a complete GitOps delivery pipeline from a single repository
How to measure and govern LLM usage with NVIDIA DCGM dashboards and OpenShift AI model metrics

Workshop Structure

Module	Lab	Topic	Duration
0 — Setup	—	Access your ROSA cluster, set up CLI credentials	20 min
1 — Platform Engineer	1.1	Deploy a GPU Machine Pool
15 min	1.2
Install Platform Operators	5 min active / ~20 min background
1.3	Discover OpenShift AI
15 min	1.4	Deploy LLM as a Service (Qwen3.6)
10 min active / ~15 min background	1.5	Explore the Developer Template	15 min
2 — Developer	2.1	Launch Your Dev Space	10 min
2.2	Build and Deploy with OpenCode	20 min
2.3	Ship It: Change → Build → Deploy	15 min
2.4	Validate Best Practices with OpenCode	10 min
3 — Observability	—	LLM Metrics & GPU Dashboards (optional)	20 min
—	—	Conclusion & Q&A	10 min

Module

Lab

Topic

Duration

0 — Setup

—

Access your ROSA cluster, set up CLI credentials

20 min

1 — Platform Engineer

1.1

Deploy a GPU Machine Pool

15 min

1.2

Install Platform Operators

5 min active / ~20 min background

1.3

Discover OpenShift AI

15 min

1.4

Deploy LLM as a Service (Qwen3.6)

10 min active / ~15 min background

1.5

Explore the Developer Template

15 min

2 — Developer

2.1

Launch Your Dev Space

10 min

2.2

Build and Deploy with OpenCode

20 min

2.3

Ship It: Change → Build → Deploy

15 min

2.4

Validate Best Practices with OpenCode

10 min

3 — Observability

—

LLM Metrics & GPU Dashboards (optional)

20 min

—

Conclusion & Q&A

10 min

Fire-and-forget timing: Long operations (GPU node provisioning, operator installation, model download) are triggered early and run in the background while you work on other labs. By the time you need the LLM, it will be ready.

Prerequisites

You will need:

A web browser
Access to the lab environment (credentials provided by your instructor)
A terminal — provided inside your Dev Space (no local install required)

Everything else — rosa, oc, kubectl, the OpenCode CLI, the Git server client — is either pre-installed in the lab environment or installed during the workshop steps.

Let’s get started!