View this email in your browser

Back Better Than Ever!
With New Format!

HI! It’s Chris here. I know it’s been some time since you’ve last heard from me here. It’s amazing how much has changed Microsoft Azure, as well as the cloud in general, since I first started Build5Nines.com and this newsletter all the way back in 2015! I’ve also had a lot of things in my personal life change in that time, as I’m sure you have too.

I’ve decided to take a new approach to the Build5Nines newsletter, and make it a bit more personal, as well as add more unique content to give you a lot more value from the newsletter than you’ve gotten in the past. Previously, Build5Nines Weekly was a link list, and that served it’s purpose at one time. Going forward I will be hand writing the Build5Nines Newsletter.

While I haven’t decided on how often I will be sending the new Build5Nines Newsletter, I will be writing to you on a regular basis. I don’t know if it’ll be weekly, bi-weekly, or monthly at this point.

Please email me at chris@build5nines.com if you have suggestions or feedback on the new Build5Nines Newsletter. I want to make it the best it can be to provide you the value you’re looking for.

As always, I appreciate your interest in Build5Nines.com and your subscription to the newsletter!

Thanks,

Chris Pietschmann

Microsoft MVP, HashiCorp Ambassador

OpenAI, GPT, and AI is everywhere! But, how does it apply for a DevOps Engineer or Site Reliability Engineer?

The latest buzz in the IT industry is AI, and more specifically GPT and other algorithms from OpenAI. The way that the new GPT and other algorithms are being exposed as what is essentially an API call makes these new advancements in AI a lot more accessible and applicable for enterprise developers. As more and more developers are adopting these AI algorithms into enterprise applications, those of us on the infrastructure side, or really anyone that fills a DevOps Engineer or SRE role is required to get up to speed on supporting the deployment and configuration of these AI algorithms.

Azure OpenAI Service with Azure Bicep and Terraform

Within Microsoft Azure, Azure OpenAI Service is a fully managed service that offers the ability to easily integrate OpenAI models (including GPT 4, GPT 3.5 Turbo, and others) into enterprise applications. The service also includes the ability to integrate the AI models with your companies own data, so the AI can answer language queries in the context of your business.

Being a fully managed service, there is support for IaC tools to provision and manage the Azure OpenAI Service using Azure Bicep, ARM Templates, and HashiCorp Terraform. In addition to provisioning the service, you can also configure AI model deployments from the IaC deployments as well.

Personally, I work a lot in both the Developer and Infrastructure spaces in my day to day consulting job. Lately, I’ve been working with Azure OpenAI Service quite a bit, and have been working to share some of my expertise in this area.

Here’s a couple blog posts on Build5Nines that take a look at using IaC to provision and configure Azure OpenAI Service:

• Azure Bicep: Create Azure OpenAI Service and GPT-4 / GPT-35-turbo Model Deployment

• Terraform: Create Azure OpenAI Service and GPT-4 / GPT-35-turbo Model Deployment

Build5Nines/AIChatUI Open Source Project

Since I’ve been working with Azure OpenAI Service from a developer perspective as well, I’ve decided to build a sample application to help demonstrate not just the IaC deployment of Azure OpenAI Service and GPT models, but also to demonstrate some basic use cases of integrating AI into your own enterprise applications.

As you’ll see within the Build5Nines/AIChatUI github project, the sample application UI is built as a static HTML page to keep it simple as possible for now. Also, the sample API backend is written with Node.js and includes two versions of the sample API implementation.

• API v1 - Is a basic implementation of integrating Azure OpenAI Service to perform chat completions with either GPT-35-turbo or GPT-4

• API v2 - Is an expanded implementation of integrating both Azure OpenAI Service  with GPT-4, and Azure AI Search service to the chat completion experience to incorporate your own enterprise data

These are the two implementations I’ve built into the project for now, and plan to expand this project out a bit more over time.

On the IaC and infrastructure deployment side of things, the project includes Azure Bicep, ARM Template, and HashiCorp Terraform templates for provisioning the Azure OpenAI Service, Azure AI Search, and GPT models. The IaC is also organized in v1 and v2 samples that apply to the UI/API v1 and v2 samples.

Below is a screenshot of the Build5Nines/AIChatUI sample application:

MVP Summit was this last week

The Microsoft MVP Summit was this past week. I learned a lot about things that are coming. However, I can’t really share anything since it’s all under NDA. I suppose I just wanted to tease you here. :)

As many of you may know, I’ve been a Microsoft MVP for a long time now. My first award was back in 2008 for Bing Maps for Enterprise, as I did a lot of mapping / GIS development work back then. For the last several years, I’ve been awarded for Microsoft Azure, and starting Build5Nines.com back in 2015 has helped me with that. I’m extremely grateful for this!

GitHub Actions: Run Pandoc to convert Markdown to Word Document

Pandoc is a great tool for document conversion and generation. I’ve written tons of training course and documentation content over the years, and have often used Pandoc to convert Markdown (.md) content to a Microsoft Word (.docx) document. This can be especially helpful since I’ve found writing text content works well with version control using Git and Markdown, then generating the required Word Document once everything is completed.

I thought I would share in this article the necessary GitHub Actions Workflow configuration to automate the use of Pandoc for document generation. This automation can help eliminate the need to have Pandoc and other potential dependencies on your local machine. The GitHub Actions Workflow can be triggered, either manually or setup to auto-trigger on commits, to easily generate the Word Document from the Markdown content in the Git repository.

Terraform: Deploy Azure App Service with Key Vault Secret Integration

One of the most popular cloud-native, PaaS (Platform as a Service) products in Microsoft Azure is Azure App Service. It enables you to easily deploy and host web and API applications in Azure. The service supports ways to configure App Settings and Connection String within the Azure App Service instance. Depending on who has access to the App Service instance, you might not want them to be able to see the values for the App Settings or Connection Strings. For this reason, Azure App Service supports integration with Azure Key Vault, a centralized and secure secrets store within Azure. This way, the application can access the secret values at runtime, but anyone with access to manage the Azure App Service instance hosting the application will not be able to see those values as they are stored in Azure Key Vault.

This article shows the steps to deploy and manage both Azure App Service and Azure Key Vault instances using HashiCorp Terraform. Then, it also shows the Terraform configuration code to both set Key Vault secrets from Terraform, as well as pass them security via input variables on the Terraform project from the DevOps deployment pipeline that orchestrates the Terraform commands.