add fixes based on kickoff meeting

antora.yml

@@ -4,6 +4,6 @@ version: 1
 nav:
 - modules/ROOT/nav.adoc
 - modules/start_here/nav.adoc
-- modules/course_guide/nav.adoc
+#- modules/course_guide/nav.adoc
 - modules/lab/nav.adoc
 - modules/references/nav.adoc

modules/ROOT/pages/index.adoc

@@ -1,34 +1,41 @@
 = Red Hat & NVIDIA AI Factory Content Accelerator
 :navtitle: Event Summary
 
-== Welcome to the Hackathon
+== The Mission: Industrialize Enterprise AI
 
-We are pleased to invite you to our AI Factory w/ NVIDIA Content Accelerator! This workshop is an intensive session designed to bridge the gap between product knowledge and real-world execution. Together, we will build the "field-opinionated" playbooks and technical blueprints required to deliver Red Hat and NVIDIA AI Factory solutions at scale.
+Welcome to the Virtual AI Factory w/ NVIDIA Content Accelerator. While we won't be in Frankfurt, this intensive, distributed hackathon is designed to bridge the gap between theoretical AI product knowledge and industrial-scale, real-world execution.
 
-== The Philosophy: Why Your Experience Matters
+In a world where AI can quickly generate code and match patterns, it still lacks the "hard knocks" of real-world trial and error. AI does not inherently understand *why* a specific architecture led to results, or why alternative methods technically work but ultimately fail in production. 
 
-In a world where AI can quickly shape and generate information, retaining the human aspect of knowledge transfer is critical. AI excels at pattern matching, but it lacks the "hard knocks" of trial and error. It does not inherently understand *why* a specific approach led to results, or why alternative methods might technically work but ultimately fail to achieve the best business outcomes.
+**That is where you come in.** Your true value lies in sharing the lessons learned through failure, iteration, and field experience. Together, we are going to translate your raw expertise into the "field-opinionated" playbooks and technical blueprints required to deliver Red Hat and NVIDIA AI Factory solutions at scale.
 
-Your true value lies in sharing the lessons learned through failure and iteration. By sharing your perspective, the nuances of your process, and the "Aha!" moments that crystallized your understanding, you translate raw blobs of information into a powerful enablement tool. You aren't just sharing a skill; you are empowering a co-worker to enable a customer and overcome the same journey in the future.
+== The Engine: Open Training & Giveback Incentives
 
-== The Open Training Framework
+To rapidly capture and distribute your playbooks, we are utilizing the **Red Hat Open Training framework** as our delivery vehicle. You will build your blueprints using a straightforward "docs-as-code" approach (AsciiDoc and Antora), allowing you to focus purely on technical accuracy while the framework handles the formatting, scaling, and publishing.
 
-To capture this knowledge, we are leveraging the Red Hat Open Training framework. Open Training is Red Hat's self-service content development model that empowers subject matter experts to share their knowledge by creating training content. This collaborative approach complements our formal development efforts, ensuring that our materials stay timely, relevant, and grounded in real-world experience. 
+**Your Impact & Reward:** By contributing to this accelerator, you are directly empowering internal associates, partners, and customers to replicate your success. Furthermore, your published blueprints are highly visible and eligible for recognition through the **Red Hat Giveback Program** (e.g., earning 4 points for new courses and 3 points for updates).
 
-By contributing to this accelerator, you help foster a culture of continuous learning at Red Hat. 
+== Hackathon Goals & Business Impact
 
-== Hackathon Goals & Outcomes
+To maintain focus and drive immediate customer value, your hackathon builds will target these core objectives. *(Note: Specific tracks and workloads will be finalized and assigned prior to our kickoff on the 15th).*
 
-To maintain focus, our work is divided into five core objectives. These will form the foundation of our new, elite content library:
+* **Production-Ready Deployments:** Develop automated Quickstarts and Playbooks (using OpenShift AI and NVIDIA NIM) that provide the field with a clear, repeatable path to production, accelerating time-to-value.
+* **Autonomous Agents:** Build enablement around the Llama Stack API and NVIDIA OpenShell runtime to guide teams in transitioning from simple chatbots to highly secure, industrial-scale AI agents.
+* **High-Performance Serving:** Establish hardware and storage benchmarking standards for NVIDIA-Certified Systems to ensure deployments meet strict AI service level objectives and reduce TCO.
+* **Security & Regulatory Compliance:** Integrate STIG-hardened containers, RHEL, and NVIDIA BlueField DPUs to deliver a zero-trust architecture for mission-critical, regulated industries.
+* **AI-First Delivery & Enablement:** Apply an AI-first delivery approach using "skills" to automate framework delivery, shifting from traditional manual consulting to scalable, AI-assisted field operations for Specialists, Consultants, and Partners.
 
-* **Production-Ready Deployments:** Develop repeatable Quickstarts, Playbooks, and Blueprints that move beyond basic training to provide a clear path for production-ready deployment.
-* **Autonomous Agents:** Build enablement on the Llama Stack API and NVIDIA OpenShell runtime to transition from simple chatbots to autonomous, industrial-scale AI agents.
-* **Performance Standards:** Establish hardware and storage benchmarking standards for NVIDIA-Certified Systems and high-performance environments.
-* **Security & Regulations:** Integrate STIG-hardened containers and RHEL for NVIDIA best practices to meet the requirements of regulated industries.
-* **Broader Field Enablement:** Create a high-impact content library to enable the broader field (Specialists, Consultants, and Partners) via existing enablement initiatives.
+== The Definition of "Done"
+
+By the end of this hackathon, a successful blueprint submission will include:
+. A completed, AI-assisted **Course Design Document (CDD)** mapping out your technical concepts to actionable tasks.
+. A working hands-on lab environment scenario (targeting the dual H100 baseline—either a step-by-step *How-To* or a troubleshooting *Break-Fix*).
+. An Antora-compliant GitHub repository containing your structured AsciiDoc pages and navigation map.
 
 == Next Steps
 
-Turning your expertise into a structured learning instrument can be challenging. In the following sections, we will outline the logistics of this event and how to use the "docs-as-code" AsciiDoc and Antora toolchain so you can start building immediately.
+Ready to start building? We have streamlined the toolchain so you can bypass the formatting hurdles and get straight to the code.
 
----
+* **Step 1:** Head over to the **xref:cdd-workflow.adoc[AI-Assisted Design Phase]** to let an LLM generate your course outline from your raw notes.
+* **Step 2:** Review the **xref:toolchain-cheatsheet.adoc[AsciiDoc & Antora Cheat Sheet]** to understand how to structure your repository.
+* **Step 3:** Start coding your playbook!

modules/lab/nav.adoc

@@ -1,3 +1,3 @@
-* Lab Infrastructure Guide
+* xref:index.adoc[Lab Infrastructure Guide]
 ** xref:rhdp.adoc[]
 // ** xref:role.adoc[]

modules/lab/pages/index.adoc

@@ -1,5 +1,45 @@
-= Lab infrastructure guide
+= NVIDIA Launchpad: Dual H100 Lab Environment
+:navtitle: Lab Environment Guide
 
-This section contains instructions on use of hands-on lab infrastructure for the training. You may either use xref:rhdp.adoc[RHDP] or xref:role.adoc[ROLE] for your lab environment.
+== The Sandbox: NVIDIA Launchpad
 
-See xref:references:faq.adoc[faq] for additional guidance.
+For this Content Accelerator, we have partnered with NVIDIA to host your hackathon workloads on **NVIDIA Launchpad**. Launchpad provides immediate, cloud-hosted access to NVIDIA-accelerated infrastructure. 
+
+Instead of burning days fighting with bare-metal provisioning or cloud quotas, you will be handed the keys to a pre-provisioned, Red Hat AI Factory-ready environment so you can focus entirely on building your blueprints.
+
+== The Hardware: Dual H100 Nodes
+
+The baseline infrastructure for your playbooks will be environments featuring **two NVIDIA H100 Tensor Core GPUs** per worker node. 
+
+**Why the H100 matters for your builds:**
+
+* **Hopper Architecture:** The H100 features a dedicated Transformer Engine, making it exceptionally fast at training and inferencing Large Language Models (LLMs) like Llama 3.
+* **Interconnect:** The dual GPUs are connected via NVLink, allowing them to pool memory and share data at incredibly high bandwidths, which is critical when deploying distributed inference engines (like vLLM or NVIDIA TensorRT-LLM) that require model sharding across multiple GPUs.
+* **Capacity:** This footprint is more than capable of hosting robust RAG (Retrieval-Augmented Generation) pipelines, vector databases, and multi-agent frameworks using the Llama Stack API.
+
+== The Software Stack: The AI Factory Floor
+
+Launchpad abstracts the hardware layer, but beneath your workloads is the full, opinionated **Red Hat AI Factory with NVIDIA** stack. Your environment will come pre-provisioned with the following:
+
+* **Base OS:** Red Hat Enterprise Linux (RHEL) CoreOS.
+* **Orchestration:** Red Hat OpenShift Container Platform (OCP).
+* **The AI Layer:** Red Hat OpenShift AI (RHOAI).
+* **The Drivers:** The **NVIDIA GPU Operator** will be pre-installed and configured. This is the critical bridge that automatically provisions the NVIDIA drivers, container toolkits, and device plugins required to expose the H100s to your Kubernetes pods.
+
+== How Access Will Work
+
+*(Note: Specific credential distribution will occur on the morning of the kickoff).*
+
+. **The Invitation:** You will receive an email invitation to the NVIDIA Launchpad portal.
+. **The Console:** Launchpad operates via a web browser. You will log in to a centralized dashboard where you can access the OpenShift web console, as well as terminal/SSH access to interact with the cluster via `oc` or `kubectl` commands.
+. **Your Namespace:** You will likely be assigned a dedicated project/namespace within the OpenShift cluster to isolate your hackathon work from other teams.
+
+== Rules of Engagement & Best Practices
+
+To ensure everyone has a successful hackathon, please observe these lab practices:
+
+WARNING: **Resource Hygiene:** A dual H100 node is incredibly powerful, but resources are still finite. When you are done testing a heavy LLM deployment, scale your pods down to zero. Do not leave idle models locking up GPU memory overnight.
+
+TIP: **Node Selectors & Tolerations:** When writing your YAML manifests or Helm charts, ensure you are using the correct node selectors and tolerations so your AI workloads actually land on the GPU-enabled worker nodes. (e.g., matching the `nvidia.com/gpu` resource limit).
+
+NOTE: **Assume Ephemerality:** Launchpad environments have a strict time-to-live (TTL) and will be destroyed after the event. Store all your AsciiDoc content, CDDs, and code manifests locally or in your assigned GitHub repository—**not** on the Launchpad cluster itself.

modules/lab/pages/rhdp.adoc

@@ -1,18 +1,51 @@
-= RHDP (skip if using ROLE)
+= Selecting a Compatible RHDP Lab Environment
+:navtitle: RHDP Lab Selection
 
-NOTE: RHDP is our primary lab development platform. 
+== From Hackathon to Production Enablement
 
-. Login to https://demo.redhat.com/catalog[RHDP,window=_blank] using your internal SSO credentials.
+During this Content Accelerator, your primary build environment is NVIDIA Launchpad, which provides massive computing power via dual H100 nodes. However, Launchpad is ephemeral. 
 
-. Identify a suitable lab environment for your course.
+For your technical blueprint to successfully scale out to Specialists, Consultants, and Partners, it must eventually be hosted on the **Red Hat Demo Platform (RHDP)**. RHDP is our primary, persistent lab development and delivery platform. 
 
-. Launch and inspect the lab environment.
+As part of your Course Design Document (CDD) process, you must identify an RHDP catalog item that can support your playbook for future learners.
 
-. Note any required changes to fit your needs.
+== The Hardware Gap: Adjusting Your Expectations
 
-. If no lab catalog suits your needs, we can create a new lab environment catalog for you.
+**Crucial Caveat:** It is highly unlikely that standard RHDP catalog items will feature dual H100 GPUs. Most cloud-provisioned GPU environments in RHDP utilize NVIDIA T4s (e.g., AWS `g4dn` series) or NVIDIA A10Gs (e.g., AWS `g5` series).
 
-For assistance, submit a request via https://docs.google.com/forms/d/e/1FAIpQLSepUaRiRdyA3PEzLP8w59reAsKRe19dL3ewpJGvJ7Gbggt-xg/viewform[this,window=_blank] form (select Open Training).
+Because these GPUs have significantly less VRAM and compute capacity than the H100 Hopper architecture, you must design your playbook with compatibility in mind:
 
-Refer to the xref:references:faq.adoc[FAQ] for more details.
+* **Model Sizing:** If your Launchpad lab uses a 70B parameter model, your RHDP equivalent may need to instruct the learner to deploy an 8B parameter model.
+* **Quantization:** You may need to add notes in your playbook about using quantized models (AWQ/GPTQ) for the RHDP environment.
+* **Performance Expectations:** Explicitly document in your blueprint that inference speeds in the training lab will be lower than the production-grade H100 sizing.
 
+== Recommended Environment: GPU-Enabled SNO
+
+For most OpenShift AI and NIM-based playbooks, we highly recommend targeting a **Single Node OpenShift (SNO)** environment equipped with GPU instances. 
+
+When searching the RHDP catalog, look for environments backed by instances such as the **AWS `g4dn.12xlarge` or `g6.12xlarge`**. 
+
+**Why this profile?**
+
+* **Resource Density:** A `12xlarge` SNO provides 48 vCPUs and 192 GiB of RAM, which is ample overhead for the OpenShift control plane, OpenShift AI operators, and your vector databases.
+* **GPU Capacity:** It provides 4x NVIDIA GPUs (T4 or A10G), which allows learners to test model sharding, multi-GPU RAG deployments, and Llama Stack multi-agent workflows without hitting immediate out-of-memory errors.
+
+== Step-by-Step: Selecting Your RHDP Lab
+
+To identify and test your long-term lab environment:
+
+. **Log In:** Navigate to link:https://demo.redhat.com/catalog[RHDP, window=_blank] and log in using your internal SSO credentials.
+. **Search the Catalog:** Search for terms like `"OpenShift AI SNO"`, `"GPU SNO"`, or `"NVIDIA"`. 
+. **Inspect the Provisioning:** Read the catalog description to ensure the NVIDIA GPU Operator is pre-installed (or note if your playbook needs to instruct the learner to install it).
+. **Launch & Test:** Provision the environment. Test your Kubernetes manifests and Helm charts to verify they successfully deploy on this hardware footprint.
+. **Document Deltas:** Note any required changes, scaling limitations, or parameter adjustments needed to make your blueprint succeed in this environment. Include these in the `# LAB INFRASTRUCTURE REQUIREMENTS` section of your CDD.
+
+== Requesting Customizations
+
+If you test the available `gx12xlarge` SNO environments and determine that your blueprint fundamentally breaks without a specific operator, configuration, or larger hardware profile, we can build it.
+
+Our Open Training and lab teams can create a new, persistent lab environment catalog specifically tailored to your blueprint. 
+
+For assistance with RHDP, or to request a new catalog build, submit a request via link:https://docs.google.com/forms/d/e/1FAIpQLSepUaRiRdyA3PEzLP8w59reAsKRe19dL3ewpJGvJ7Gbggt-xg/viewform[this intake form, window=_blank] (ensure you select **Open Training** from the dropdown menu).
+
+NOTE: For more details on lab support and platform differences, refer to the xref:references:faq.adoc[FAQ].

modules/references/pages/asciidocqrg.adoc

@@ -1,3 +1,136 @@
+= AsciiDoc & Antora Toolchain Cheat Sheet
+:navtitle: Toolchain Cheat Sheet
+
+== The Goal: Content Over Formatting
+When building your AI Factory blueprints, we want you focused on the architecture, not fighting with markup. 
+
+We use **Antora** to build the site and **AsciiDoc** for the content. It functions as "docs-as-code." Use this cheat sheet for the 90% of formatting you will actually need during the hackathon.
+
+---
+
+== 1. The Antora Repository Map
+When you clone your assigned GitHub repository, you will see a specific folder structure. Do not fight the structure; Antora requires it to compile your course.
+
+[source,text]
+----
+repository-root/
+├── antora.yml               <-- The Orchestrator (Defines course title & module links)
+└── modules/
+    └── chapter1/
+        ├── nav.adoc         <-- The Map (The table of contents for this chapter)
+        ├── pages/           <-- YOUR CONTENT: All your .adoc files go here!
+        └── images/          <-- YOUR MEDIA: Architecture diagrams and screenshots go here!
+----
+
+**Golden Rule:** If you write a new `.adoc` file in the `pages/` directory, you **must** add it to the `nav.adoc` file so it shows up in the course menu.
+*Example `nav.adoc` entry:* `** xref:my-new-page.adoc[Setting up the GPU Operator]`
+
+---
+
+== 2. AsciiDoc Essentials
+
+=== Headings
+Keep it simple. Use equal signs. (Remember: `=` is reserved for the document title at the very top of the page).
+
+[source,asciidoc]
+----
+== Level 1 Section (Heading 2)
+=== Level 2 Section (Heading 3)
+==== Level 3 Section (Heading 4)
+----
+
+=== Text Formatting
+[source,asciidoc]
+----
+Make text **bold** or *italicized*.
+For inline code, commands, or parameters, use the plus sign: +oc get pods+
+----
+
+=== Lists
+[source,asciidoc]
+----
+* First bullet point
+** Nested bullet point
+
+. Step one (Use a single period for numbered lists)
+. Step two
+.. Nested step
+----
+
+---
+
+== 3. Code Blocks & Terminal Output
+Since you are building technical playbooks, you will use code blocks constantly. Always declare the source language for proper syntax highlighting.
+
+**For YAML/Config files:**
+[source,asciidoc]
+......
+[source,yaml]
+----
+apiVersion: v1
+kind: Namespace
+metadata:
+  name: nvidia-gpu-operator
+----
+......
+
+**For Terminal/Bash commands:**
+[source,asciidoc]
+......
+[source,bash]
+----
+curl -X POST http://llama-stack-api:8080/v1/chat/completions \
+  -H "Content-Type: application/json" \
+  -d '{"model": "llama3-8b", "messages": [{"role": "user", "content": "Hello"}]}'
+----
+......
+
+---
+
+== 4. Security Warnings & Pro-Tips (Admonitions)
+Break up walls of text and draw attention to critical requirements (like STIG compliance or destructive commands) using visual callouts.
+
+[source,asciidoc]
+----
+NOTE: The NVIDIA GPU Operator must be installed before proceeding.
+
+WARNING: **STIG Mandate:** Ensure that all containers are configured to run as non-root.
+
+TIP: If you are benchmarking storage, consider using the local NVMe drives attached to the A100 nodes.
+----
+
+---
+
+== 5. Images and Links
+
+**To embed an architecture diagram:**
+(Ensure the image file is actually placed in your `images/` directory!)
+[source,asciidoc]
+----
+image::doca-argus-telemetry.png[DOCA Argus Architecture Diagram]
+----
+
+**To link to an external site:**
+(Always use `window=_blank` so the learner doesn't lose their place in the course!)
+[source,asciidoc]
+----
+Read the link:https://docs.nvidia.com[NVIDIA Documentation, window=_blank].
+----
+
+**To link to another page in your repository:**
+[source,asciidoc]
+----
+Check out the xref:troubleshooting-rayclusters.adoc[Troubleshooting Guide].
+----
+
+---
+
+== Ready? Set. Build.
+You have your CDD outline, you understand the repository structure, and you know the AsciiDoc syntax. 
+
+**It's time to start coding your blueprint.**
+
+////
 = AsciiDoc Quick Reference Guide
 :navtitle: AsciiDoc Reference
 
@@ -244,4 +377,6 @@ video::video-file.mp4[]
 [source,asciidoc]
 ----
 audio::audio-file.wav[] 
-----
+----
+
+////