ABOUT THE SERIES

MLconf interviews experts from Red Hat about three key aspects of MLOps: data engineering, initial model deployment, and model governance.

Episode 1: Data Engineering

Hema Veeradhi, Senior Software Engineer with Red Hat, shares her perspective about data engineering in the context of MLOps, including:

• How to get started with data engineering;
• How to address challenges with acquiring and extracting data;
• What considerations are necessary to scope processing requirements for exploring and preparing data when dealing with data-hungry algorithms; and
• What data scientists and developers should have in place with regard to data engineering before deploying models.

 

Episode 2: Initial Model Deployment

Trevor Royer, Senior Consultant with Red Hat, shares his thoughts about model deployment, including:

• How to get started with deploying models;
• What options are available for consuming models after they are deployed;
• What considerations to keep in mind with regard to continuous integration and continuous delivery (CI/CD) of deployed models;
• How to determine if a model deployment is successful;
• How development teams and data science teams collaborate effectively; and
• What organizations can do to align development with data science.

 

Episode 3: Model Governance

Audrey Reznik, Senior Principal Software Engineer with Red Hat, shares her perspective about model governance in the context of MLOps, including:

• What model governance entails, as well as where data governance and process governance fit in;
• What criteria concerning security and controls are essential for model governance;
• Which best-practice requirements, including algorithmic reproducibility, facilitate model governance;
• What risks organizations are most likely to encounter when introducing algorithms from public or external sources, as well as how to mitigate these risks; and
• Which aspects of model governance MLOps practitioners need to keep in mind.