ZenML: Your Open-Source Path Forward After cnvrg.io

The MLOps landscape is evolving rapidly, and with change comes opportunity. Intel's recent announcement that cnvrg.io (Tiber AI Studio) will be discontinued by August 2025 has left many data science teams evaluating their options. If you're among those affected by this transition, we want to share how ZenML can provide a sustainable, open-source path forward for your MLOps journey.

Understanding What's at Stake

cnvrg.io earned its place in many organizations by providing a comprehensive, unified platform for ML workflows. Companies like Seagate, Monday.com, and Wargaming built robust ML pipelines using cnvrg.io, leveraging its ability to handle hybrid infrastructure and orchestrate end-to-end MLOps processes. The platform's collaborative features and integrated tooling helped teams deliver AI projects efficiently.

However, the upcoming end-of-life presents more than just a challenge – it's an opportunity to rethink your MLOps strategy. Rather than rushing to another proprietary platform and potentially facing similar vendor lock-in issues down the road, now is the perfect time to consider how an open-source approach with ZenML could provide greater flexibility and long-term stability.

The Open-Source Advantage

At its core, ZenML represents a fundamentally different approach to MLOps. As an open-source framework released under the Apache 2.0 license, ZenML puts control back in your hands. This isn't just about avoiding vendor lock-in – though that's certainly important given recent events. It's about building on a foundation that can't be taken away or discontinued. The codebase is transparent, extensible, and shaped by a growing community of practitioners who understand the real-world challenges of deploying ML systems.

Bridging Familiar Territory

If you're coming from cnvrg.io, you'll find that ZenML supports many of the workflows you've built, but with an important twist. Take pipeline orchestration, for instance. Where cnvrg.io provided a unified but closed system, ZenML offers code-first pipeline definitions that are version-controlled and portable. Your pipelines become part of your codebase, versioned in Git alongside your model code, making them easier to review, share, and maintain.

Infrastructure management takes on a new dimension with ZenML. Similar to cnvrg.io's hybrid capabilities, ZenML works seamlessly across cloud and on-premises environments. But rather than locking you into a proprietary scheduler, ZenML's infrastructure-agnostic design lets you leverage your preferred orchestration tools. Whether you're running on Kubeflow, Airflow, or another orchestrator, ZenML adapts to your stack, not the other way around.

For experiment tracking and model management, ZenML takes an integration-first approach. Instead of forcing you into a monolithic system, ZenML connects with best-in-class tools like MLflow, Weights & Biases, and others. This means you can choose the right tool for each job while maintaining a consistent pipeline interface. Full experiment reproducibility comes built-in through ZenML's pipeline versioning system, ensuring you can track and reproduce your ML workflows with confidence.

A Practical Path Forward

We understand that migrating from cnvrg.io is no small undertaking. That's why we've designed ZenML to support a gradual transition. You can start with a single project or use case, running ZenML alongside your existing cnvrg.io implementation as you migrate pipelines one at a time. This measured approach helps maintain production stability while you build confidence in your new stack.

What's more, ZenML is designed to work with your existing infrastructure. Keep your Kubernetes clusters, cloud services, and other resources – ZenML integrates with them rather than requiring a complete platform overhaul. This pragmatic approach saves both time and resources during the transition.

For teams accustomed to cnvrg.io's collaborative features, ZenML offers a different but powerful paradigm. The code-first approach resonates naturally with ML engineers and data scientists, while Git integration provides a robust foundation for collaboration. For teams wanting more platform-like features, ZenML Cloud offers managed services that can smooth the transition from cnvrg.io's unified interface.

From Theory to Practice

Let's look at how a typical cnvrg.io workflow translates to ZenML in practice. Consider this simple training pipeline:

from zenml import pipeline, step

@step
def data_ingestion():
    # Your existing data loading code
    return df

@step
def model_training(data):
    # Your existing training code
    return model

@pipeline
def training_pipeline():
    data = data_ingestion()
    model = model_training(data)

# Run it locally or on any infrastructure
training_pipeline()

This straightforward example demonstrates how ZenML maintains familiar concepts while providing more flexibility in implementation. Your existing ML code can be wrapped in ZenML steps and composed into pipelines that run anywhere – from your local machine to production clusters.

Building for the Future

Choosing ZenML isn't just about replacing cnvrg.io – it's about building a more sustainable MLOps foundation. The extensible nature of ZenML means you can add new tools and capabilities as your needs evolve, create custom integrations, and even contribute back to the open-source community. This flexibility extends to cost management as well; you'll pay for compute resources rather than platform licenses, scaling based on actual needs rather than seat licenses.

The ZenML community provides another crucial advantage. Active open-source development means regular updates and improvements driven by real-world use cases. Commercial support through ZenML Cloud gives enterprises the backing they need, while the open-source community offers a wealth of knowledge and experience.

Taking the First Step

The path forward from cnvrg.io to ZenML is clear and achievable. Start by exploring our quickstart guide at https://docs.zenml.io/user-guide/starter-guide. Join our thriving Slack community at zenml.io/slack, where you'll find others making similar transitions and experts ready to help. For personalized guidance, book a consultation at https://www.zenml.io/book-your-demo to discuss your specific needs and migration strategy.

With the August 2025 deadline approaching, now is the perfect time to evaluate your options. ZenML offers a sustainable, open-source path forward that puts you in control of your MLOps destiny. We're here to help you make that transition successfully.

Visit our documentation at docs.zenml.io and explore our GitHub repository at github.com/zenml-io/zenml to learn more about how ZenML can support your MLOps journey.