Kubeflow

Kubernetes-based Machine Learning Platform

Kubeflow automates repetitive work in the overall pipeline for development, learning, and deployment processes of the machine learning model. MLOps1) environments based on the machine learning platform are provided, offering integrated management of learning data and models as well as operational data.
1) MLOps : A ML engineering discipline that aims to unify machine learning development (Dev) and machine learning system operation (Ops)

Overview

  • Cloud Native MLOps Environments

    Kubeflow offers ML model development environments optimized for cloud, enabling Kubernetes-based linking with various open source software.

  • Usage Convenience for Big Data

    The standardized environments support a range of machine learning frameworks from TensorFlow, PyTorch, scikit-learn, and Keras. The pipeline for the entire development, learning and deployment processes of machine learning models are automated to ensure simple configuration/creation as well as reuse of the models.

  • Add-on Features

    Expanded open source Kubeflow from distributed learning job execution/monitoring to inference service management/analysis and job queue management are provided on Samsung Cloud Platform. Users can also enjoy job schedulers (FIFO, Bin-packing, and Gang-based), GPU fraction, GPU resource monitoring, Kubeflow engine logging and more add-on features that are usually unavailable on open source software.

01

04

Service Architecture

  1. User
  2. Console
  3. Kubernetes Engine : CPU Worker Nodes, GPU Worker Nodes, Persistent Volume
  1. Data Scientist, MLOps Engineer
  2. Kubeflow : Pipeline, Meta data, Model Serving, Notebook, Hyper Parameter Tuning ...
  3. Jupyter Notebook → Model Development → Model Training → Hyper Para.Tuning → Model Serving → Inference Application.

Key Features

  • Basic function (Both Mini and Enterprise)

    - Create AI platform (auto-deployment/configuration), view (platform version, resource status), and delete
    - Provide Jupyter Notebook : Model development, learning, inference
    - Automate machine learning pipeline workflow
    - Provide other open source Kubeflow default feature

  • Additional feature (Enterprise)

    - Advanced AI/ML platform dashboard
    - AI/ML notebook server : Base image, user-defined image
    - AI/ML job : Job creation, template, archive, scheduling, execution, monitoring
      ※ Support GPU resource monitoring, GPU fraction
    - Build and manage user image
    - AI JumpStarter and ETM (Experiment Tracking Management)
    - Serving : Dashboard, register/manage model, inference, predictions, and visualization
    - Managing platform resource : Manage resource usage by project, monitor resource usage
    - Manage project user/permissions, admin feature, adjust platform configuration

Pricing

    • Offering
    • Samsung Cloud Platform Kubeflow SW packaging
    • Billing
    • Charged by the hour for the scale and usage of deployed Kubeflow
      ※ Samsung Cloud Platform for user environment configuration charged additionally
Let’s talk

Whether you’re looking for a specific business solution or just need some questions answered, we’re here to help

Share KakaoTalk Facebook Twitter Linkedin Close Share Box