Get in Touch
 Duration 14 hours

Course Outline

Kubeflow Introduction

  • Comprehending Kubeflow’s objectives and structural design
  • Overview of primary components and ecosystem tools
  • Deployment strategies and platform functionalities

Interacting with the Kubeflow Dashboard

  • Navigating the user interface
  • Administration of notebooks and workspaces
  • Integration of storage systems and data sources

Foundations of Kubeflow Pipelines

  • Architecture and design of pipeline components
  • Developing pipelines using the Python SDK
  • Execution, scheduling, and monitoring of pipeline instances

Training ML Models via Kubeflow

  • Strategies for distributed training
  • Leveraging operators such as TFJob, PyTorchJob, and others
  • Resource optimization and autoscaling within Kubernetes

Model Serving through Kubeflow

  • Introduction to KFServing / KServe
  • Model deployment with bespoke runtimes
  • Oversight of revisions, scaling behaviors, and traffic distribution

Oversight of ML Workflows on Kubernetes

  • Version control for data, models, and artifacts
  • Integration of CI/CD processes for ML pipelines
  • Security protocols and role-based access management

Production ML Best Practices

  • Architecting dependable workflow structures
  • Implementation of observability and monitoring tools
  • Resolution of frequent Kubeflow challenges

Advanced Subjects (Optional)

  • Multi-tenant Kubeflow configurations
  • Hybrid and multi-cluster deployment contexts
  • Extending Kubeflow via custom components

Conclusion and Future Steps

Requirements

  • A grasp of containerized application architectures
  • Proficiency with fundamental command-line operations
  • Basic knowledge of Kubernetes principles

Target Audience

  • ML engineers
  • Data scientists
  • DevOps teams unfamiliar with Kubeflow

Number of participants


Price per participant (excl. VAT)

Testimonials (3)

Upcoming Courses

Related Categories