Added second doc: Run an LLM workflow with Flyte

Author: schaffererin

articles/aks/TOC.yml

@@ -848,6 +848,8 @@
           href: ai-toolchain-operator.md
         - name: Deploy data and ML pipelines with Flyte
           href: use-flyte.md
+        - name: Run an LLM workflow with Flyte
+          href: llm-workflow-flyte.md
     - name: DevOps
       items:
         - name: Azure DevOps Project

articles/aks/llm-workflow-flyte.md

@@ -0,0 +1,79 @@
+---
+title: Run an LLM inferecing workflow with Flyte on Azure Kubernetes Service (AKS)
+titleSuffix: Azure Kubernetes Service
+description: Learn how to run an LLM inferencing workflow with Flyte on Azure Kubernetes Service (AKS).
+ms.topic: how-to
+ms.date: 05/22/2024
+author: schaffererin
+ms.author: schaffererin
+ms.service: azure-kubernetes-service
+---
+
+# Run an LLM inferencing workflow with Flyte on Azure Kubernetes Service (AKS)
+
+This article shows you how to run an LLM inferencing workflow with [Flyte](https://docs.flyte.org/en/latest/introduction.html) on Azure Kubernetes Service (AKS).
+
+In this article, you learn how to:
+
+> [!div class="checklist"]
+>
+> * Install the PyTorch plugin for Flyte.
+> * Create a Flyte task that uses the PyTorch Lightning deep learning framework to define LLM inferencing logic.
+> * Create a Flyte workflow that orchestrates the LLM inferencing task.
+> * Build Docker images for the task and workflow.
+> * Register the workflow with Flyte.
+> * Run the workflow on AKS.
+
+## Before you begin
+
+* See [Build and deploy data and machine learning pipelines with Flyte on Azure Kubernetes Service (AKS)](./use-flyte.md) for **initial setup and prerequisites**.
+* This article assumes a basic understanding of Kubernetes concepts. For more information, see [Core Kubernetes concepts for Azure Kubernetes Service (AKS)](./concepts-clusters-workloads.md).
+
+## Install the PyTorch plugin for Flyte
+
+The PyTorch plugin for Flyte leverages the Kubeflow training operator to provide a streamlined interface for conducting distributed training. For more information, see [PyTorch Distributed plugin for Flyte](https://docs.flyte.org/en/latest/flytesnacks/examples/kfpytorch_plugin/index.html#).
+
+* Install the PyTorch plugin for Flyte using the following `pip` command:
+
+    ```bash
+    pip install flytekitplugins-kfpytorch
+    ```
+
+## Create a Flyte task
+
+A Flyte *task* is a function that performs a specific unit of work. Tasks are assembled into workflows to create data and machine learning pipelines. When deployed to a Flyte cluster, each task runs in its own Pod on the cluster. For more information, see [Flyte tasks](https://docs.flyte.org/en/latest/user_guide/basics/tasks.html).
+
+Add step(s)
+
+## Create a Flyte workflow
+
+Flyte *workflows* consist of multiple tasks or workflows to produce a desired output. For more information, see [Flyte workflows](https://docs.flyte.org/en/latest/user_guide/basics/workflows.html).
+
+Add step(s)
+
+## Build Docker images
+
+Add step(s)
+
+## Register the workflow with Flyte
+
+Packaging and registering, or *deploying*, a workflow with Flyte enables you to scale, schedule, and manage your workloads. There are different methods for registering your workflows with Flyte. In this article, we iterate on a single workflow script. For more information, see [Registering Flyte workflows](https://docs.flyte.org/en/latest/flyte_fundamentals/registering_workflows.html).
+
+* Make sure you're in your Flyte project directory and register your workflow using the `pyflyte run` command.
+
+    ```bash
+    pyflyte run XYZ
+    ```
+
+    Your output should look similar to the following example output:
+
+    ```output
+    XYZ
+    ```
+
+## Run your workflow
+
+
+
+## Next steps
+

articles/aks/use-flyte.md

@@ -3,24 +3,27 @@ title: Build and deploy data and machine learning pipelines with Flyte on Azure
 titleSuffix: Azure Kubernetes Service
 description: Learn about Flyte, an open-source platform for building and deploying data and machine learning pipelines on Azure Kubernetes Service (AKS).
 ms.topic: how-to
-ms.date: 05/21/2024
+ms.date: 05/22/2024
 author: schaffererin
 ms.author: schaffererin
 ms.service: azure-kubernetes-service
 ---
 
 # Build and deploy data and machine learning pipelines with Flyte on Azure Kubernetes Service (AKS)
 
-This article shows you how to use Flyte on Azure Kubernetes Service (AKS). Flyte is an open-source workflow orchestrator that unifies machine learning, data engineering, and data analytics stacks to help you build robust and reliable applications.
+This article shows you how to use Flyte on Azure Kubernetes Service (AKS). Flyte is an open-source workflow orchestrator that unifies machine learning, data engineering, and data analytics stacks to help you build robust and reliable applications. When using Flyte as a Kubernetes-native workflow automation tool, you can focus on experimentation and providing business value without increasing your scope to infrastructure and resource management.
 
 For more information, see [Introduction to Flyte](https://docs.flyte.org/en/latest/introduction.html).
 
 ## Prerequisites
 
 * An Azure subscription. If you don't have an Azure subscription, you can create a [free account](https://azure.microsoft.com/free).
+  * If you have multiple subscriptions, make sure you select the correct one using the `az account set --subscription <subscription-id>` command.
 * The Azure CLI installed and configured. Check your version using the `az --version` command. If you need to install or upgrade, see [Install the Azure CLI](https://docs.microsoft.com/cli/azure/install-azure-cli).
 * The Helm CLI installed and updated. Check your version using the `helm version` command. If you need to install or upgrade, see [Install Helm](https://helm.sh/docs/intro/install/).
 * The `kubectl` CLI installed and updated. Install it locally using the `az aks install-cli` command or using [Install kubectl](https://kubernetes.io/docs/tasks/tools/install-kubectl/).
+* A local Docker development environment. For more information, see [Get Docker](https://docs.docker.com/get-docker/).
+* `flytekit` and `flytectl` installed. For more information, see [Flyte installation](https://flyte-next.readthedocs.io/en/latest/introduction.html#installation).
 
 > [!NOTE]
 > If you're using the Azure Cloud Shell, the Azure CLI, Helm, and kubectl are already installed.
@@ -125,9 +128,9 @@ For more information, see [Introduction to Flyte](https://docs.flyte.org/en/late
 
 In this article, you learned how to deploy a Flyte chart on AKS. To learn more about deployments on AKS, see the following articles:
 
-* [Deploy an application that uses OpenAI on Azure Kubernetes Service (AKS)](./open-ai-quickstart.md)
+* [Run an LLM inferencing workflow with Flyte on Azure Kubernetes Service (AKS)](./llm-workflow-flyte.md)
 * [Install existing applications with Helm on Azure Kubernetes Service (AKS)](./kubernetes-helm.md)
-* [Deploy a containerized application to Azure Kubernetes Service (AKS)](./tutorial-kubernetes-deploy-application.md
+* [Deploy a containerized application to Azure Kubernetes Service (AKS)](./tutorial-kubernetes-deploy-application.md)
 
 <!-- LINKS -->
 [az-group-create]: /cli/azure/group#az-group-create