• Online
  • Webcast

ML Ops: High Quality ML Model Deployments, Faster


  • Webcast


  • ML Ops
  • DevOps
  • Azure Machine Learning

As machine learning and data science approaches are increasingly being used to make core business decisions, we rely on the accuracy of the models and their availability to decision-makers.

ML Ops is an approach to enable data scientists to collaborate on models, track and audit the model-building process, and streamline deployment to deliver value to the business more quickly and reliably.

In this session, Thorogood consultants Andrew Kennedy and Ojaswi Kumar introduce ML Ops, discussing how it builds on best-practice approaches from DevOps for software engineering, and update it for the Data Science and Machine Learning project lifecycle. Starting with model tracking and management, they look at how you can increase visibility across the model-building stage, allowing for a more collaborative approach while gaining better insight into the many models deployed across the organization.

They also discuss how continuous integration and continuous deployment apply to machine learning models and consider when and how to use automated testing to increase confidence that the right models are deployed successfully. Beyond the standard ML Ops blueprint, they look at specific industry scenarios to consider which ML Ops practices make sense for you to adopt in your projects.

ML Ops: High Quality ML Model Deployments, Faster

In this recorded webcast we:

  • Demonstrate how Azure Machine Learning and Azure DevOps combine to provide a technical approach to ML Ops.
  • Show the range of features included in Azure ML for training and tracking and evaluate how you can use Pipelines, Repos and Testing in Azure DevOps to support the deployment cycle for Data Science.

Is it for you?

  • Are you building advanced analytics solutions incorporating Machine Learning and AI and are evaluating the best ways to handle your project lifecycle?
  • Are you looking at increasing collaboration between your data teams, increase the reliability of your solutions and ease the process of your Machine learning model deployments?