MLOps Workflow

DKube supports an end-to-end MLOps workflow from feature engineering through production deployment. The platform is based on the popular Kubeflow framework, bringing together its powerful components and enhancing them with best-in-class capabilities such as:

  • Feature Engineering
  • Tracking and Lineage
  • MLFlow-based metric collection and compare
  • Flexible data source integration
  • CI/CD-based automation
kubeflow

This is all integrated into a flexible, UI-based workflow. It is intuitive enough to allow team members to be collaborating on real research within hours of starting the installation.

DKube is optimized for on-prem installation out-of-the-box. The difficult, time-consuming task of integrating the hardware with the software components is handled by our Helm-based installation. Because it runs on top of Kubernetes, DKube works with the same look, feel, workflow, and reliability on a cloud-based platform. And your work can be quickly and easily migrated back and forth.

DKube is standards-based from the ground up. It uses the best-in-class frameworks and tools, including:

  • TensorFlow
  • PyTorch
  • Scikit Learn
  • Keras
  • JupyterLab
  • RStudio
  • Katib
logos

And it supports the most common authorization standards: GitHub & LDAP.

Flexible code and data integration is built into the workflow. It supports:

  • The most popular code repositories, including GitHub, GitLab, and Bitbucket
  • The most common storage standards for data and models, including GitHub, GitLab, Bitbucket, AWS S3, Minio, Google Cloud Storage, and Redshift.

Learn More