Preview Mode Links will not work in preview mode

Azure DevOps Podcast


Message from Jeffrey Palermo: Howdy.  Welcome to my podcast.  I hope it helps you ship software more quickly and more reliably. Through the topics and guests, I hope your life is made easier. Let me know what topics would be helpful for you.

Podcast sponsor: Clear Measure is a software engineering firm and Microsoft Gold Partner empowering development teams to be their best. Clear Measure equips developers with the devops tools, methods, and automation necessary to focus on building their applications rather than wrestling with builds, deployments, or environments. Click clear-measure.com to see whether a devops implementation is right for you.

Oct 29, 2018

This week, your host, Jeffrey Palermo, interviews Damian Brady. Damian is a Senior Cloud DevOps Developer Advocate at Microsoft, helping customers implement DevOps methods on the Microsoft platform. He’s been with Microsoft for just over a year now and formerly served as a developer for Octopus Deploy.

 

In this episode, Damian and Jeffrey talk all things data science and machine learning. Damian answers key questions such as: what has been the biggest change in the area of data science since the Azure DevOps release? What does source control look like for data science projects in DevOps? And more. He also explains some of the interesting architectures he has put together for machine learning and walks Jeffrey through the process of his machine learning model from source control, building, packaging, and finally, to deploying. He also gives his recommendations for those who want to go even further with data science after listening to this week’s episode.

 

Topics of Discussion:

[:52] About today’s guest, Damian Brady.

[1:06] Damian introduces himself and explains his role at Microsoft.

[1:46] Which group Damian is presently on at Microsoft.

[4:14] With the Azure DevOps release, what’s the big change in the area of data science? What is going to be different for people building or running models?

[6:47] For data science projects what does the source control look like?

[8:49] For the Microsoft ML, is there a particular format that the data is stored in, in source control?

[9:09] If the data is large and needs to be versioned, what are the current methods people are using?

[11:06] A word from Azure DevOps sponsor: Clear Measure.

[11:39] Some of the interesting architectures Damian has put together for machine learning.

[16:10] Damian walks Jeffrey through his machine learning model from source control to building, to packaging up the release, to deploying.

[19:20] For this type of model, where would be the physical environment where it’s measuring information?

[20:24] Damian talks firewall rules, permissions, and security.

[23:16] The advantages of using Azure’s IoT Hub.

[24:46] Damian talks about the new open source features that were added with the release.

[28:20] Does Damian still encounter customers who say they don’t want to use Microsoft products because they don’t realize they’re open source?

[29:36] Is it true that VS Code is the most popular editor?

[31:03] One of the huge advantages of using open source.

[31:53] Damian talks build agents.

[33:33] About the new Windows-hosted container build agent.

[35:50] Damian’s recommendation for listeners who want to go further with data science after listening to this week’s podcast!

 

Mentioned in this Episode:

Azure DevOps

Azure Pipelines

Octopus Deploy

Clear Measure (Sponsor)

Buck Hodges on the introduction to Azure DevOps Services - Episode 001

Donovan Brown on How to Use Azure DevOps Services - Episode 002
Source control in Azure DevOps

Ubuntu

Machine Learning (ML)

Amazon Web Services (AWS)

Azure Cognitive Services

CustomVision.ai

Raspberry Pi
Azure Data Center

.NET Core

Python

GitHub

Azure IoT Hub

ADP Summit
VS Code

Docker Compose

Subversion

Chocolatey

 

Want to Learn More?

Visit AzureDevOps.Show for show notes and additional episodes.

 

Follow Up with Our Guest:

Damian Brady’s LinkedIn