Canonical updates Charmed Kubeflow with MLFlow integration

Canonical, the U.K.-based publisher of Ubuntu, has announced the latest version of its free-to-use open-source MLOps platform Charmed Kubeflow. The release, available as version 1.4, brings major usability changes, including MLFlow integration, to enhance how enterprises collaborate on AI and machine learning innovation throughout the lifecycle of the project.

Designed to run on top of Kubernetes, Charmed Kubeflow enables data scientists to develop and deploy machine learning algorithms and workflows. It supports multiple tools for machine learning (ML) – including Argo, PyTorch, and TensorFlow – and is suitable for all AI use-cases, ranging from developing AI algorithms on a local machine to training models on a public or private cloud.

Charmed Kubeflow 1.4 update

With its latest release, the browser-based solution is getting support to integrate MLFlow, an open source platform for advanced AI/ML model lifecycle management. MLFlow comes with a centralized model registry and features for experimentation, reproducibility, and deployment. It can enable truly automated model lifecycle management with the ability to detect model drift and trigger a Kubeflow model retraining pipeline.

In addition to this, the update brings a unified training operator supporting TensorFlow, MXNet, XGBoost, and PyTorch. This, as the company explains, will further simplify the product, improving future extensibility and ensuring the consumption of fewer resources on the Kubernetes cluster. It also offers full support for multi-user deployment scenarios out of the box for all Kubeflow components, including Kubeflow notebooks, pipelines, and experiments. 

This, Canonical claims, will help improve governance and reduce the occurrence of shadow-IT environments, whilst helping to combat organizational data leakage.

Data scientists can begin using Charmed Kubeflow 1.4 with Juju, the unified operator framework for hyper-automated management of applications running on both virtual machines and Kubernetes. The new release can also be deployed to any conformant Kubernetes cluster using a single Juju command. Charmed Kubeflow also comes with 24/7 support and fully-managed service options from Canonical.

According to Cognilytica, the market for MLOps solutions is expected to grow from $350 million to $4 billion by 2025.

Source: https://venturebeat.com/2022/01/25/canonical-updates-charmed-kubeflow-with-mlflow-integration/