ZenML Blog

As our AI/ML projects evolve and mature, our processes and tooling also need to keep up with the growing demand for automation, quality and performance. But how can we possibly reconcile our need for flexibility with the overwhelming complexity of a continuously evolving ecosystem of tools and technologies? MLOps frameworks promise to deliver the ideal balance between flexibility, usability and maintainability, but not all MLOps frameworks are created equal. In this post, I take a critical look at what makes an MLOps framework worth using and what you should expect from one.

Connecting model training pipelines to deploying models in production is seen as a difficult milestone on the way to achieving MLOps maturity for an organization. ZenML rises to the challenge and introduces a novel approach to continuous model deployment that renders a smooth transition from experimentation to production.

This week I spoke with Kush Varshney, author of 'Trustworthy Machine Learning', a fantastic guide and overview of all of the different ways machine learning can go wrong and an optimistic take on how to think about addressing those issues.

ML practitioners today are embracing data-centric machine learning, because of its substantive effect on MLOps practices. In this article, we take a brief excursion into how data-centric machine learning is fuelling MLOps best practices, and why you should care about this change.

This week I spoke with Matt Squire, the CTO and co-founder of Fuzzy Labs, where they help partner organizations think through how best to productionise their machine learning workflows.

With ZenML 0.6.3, you can now run your ZenML steps on Sagemaker, Vertex AI, and AzureML! It’s normal to have certain steps that require specific infrastructure (e.g. a GPU-enabled environment) on which to run model training, and Step Operators give you the power to switch out infrastructure for individual steps to support this.

This week I spoke with Emmanuel Ameisen, a data scientist and ML engineer currently based at Stripe. Emmanuel also wrote an excellent O'Reilly book called 'Building Machine Learning Powered Applications', a book I find myself often returning to for inspiration and that I was pleased to get the chance to reread in preparation for our discussion.

As we outgrew our initial template Github Action workflow, here's the five things we added to our Github Action arsenal to fit our growing needs: Caching, Reusable Workflows, Composite Actions, Comment Triggers and Concurrency Management.

An exploration of some frameworks created by Google and Microsoft that can help think through improvements to how machine learning models get developed and deployed in production.