Introduction to Kubeflow (en)
The course ‘Introduction to Kubeflow’ provides a comprehensive guide on installing, utilizing, and customizing Kubeflow, an open-source framework designed to simplify the development, training, and deployment of machine learning models on Kubernetes. Throughout the course, students will be introduced to the basic concepts of Kubeflow and the installation process in a Kubernetes cluster. Subsequently, they will be guided through interacting with Kubeflow’s Notebook server for the development and execution of machine learning models. The course will also cover the creation of custom images for running models on Kubeflow, as well as the use of KServe for model deployment in production environments.
Additionally, students will explore the Discovery Pipeline and experiments for managing the machine learning workflow and optimizing hyperparameters. The course will culminate with a practical demonstration on customizing the Kubeflow dashboard and managing the entire ML and MLOps lifecycle. In summary, ‘Introduction to Kubeflow’ offers students an in-depth overview of Kubeflow’s capabilities and functionalities, preparing them to utilize this powerful tool in implementing and managing machine learning projects on Kubernetes.
CODE: DSAI202
Category: Artificial Intelligence
Teaching methodology
The course includes educational laboratories in which each student will be able to work in order to complete training exercises that will provide practical experience in using the instrument, for each of the topics covered during the course.
Prerequisites
- Basic knowledge of containerization concepts, particularly Docker.
- Familiarity with the Linux/Unix development environment and command line.
- Experience with provisioning and managing cloud resources, such as Google Cloud Platform or Kubernetes.
- Understanding of machine learning fundamentals and concepts of machine learning model development and deployment (MLOps).
- Familiarity with Python and the machine learning libraries ecosystem.
The following is an overview of course content:
Install Kubeflow: Setting up Kubeflow, an open-source framework designed to simplify the development, training, and deployment of machine learning models on Kubernetes clusters.
Workbenches: Exploring the workbenches within Kubeflow, providing environments for various tasks such as data exploration, model development, and experimentation.
Custom Images: Creating and utilizing custom Docker images for running machine learning models on Kubeflow, allowing for flexibility and customization.
KServe: Introduction to KServe, a Kubernetes-native server for machine learning inference, enabling scalable and reliable deployment of models in production environments.
Pipelines: Understanding and utilizing Kubeflow Pipelines for building and managing end-to-end machine learning workflows, facilitating automation and reproducibility.
AutoML: Exploring the AutoML capabilities within Kubeflow, which automate the process of model selection, hyperparameter tuning, and feature engineering.
Explore Dashboard: Navigating and customizing the Kubeflow dashboard to visualize and monitor machine learning experiments, models, and resources.
MLOps Concepts: Delving into MLOps (Machine Learning Operations) concepts, focusing on best practices for managing the entire machine learning lifecycle, including development, deployment, and monitoring.
At the end of the course, participants will be able to:
- Install and configure Kubeflow.
- Interact with the notebook server.
- Create and use custom images.
- Experiment with KServe.
- Utilize the discovery pipeline and experiments.
- Optimize hyperparameters.
- Customize the dashboardÂ
- Understand fundamental concepts of Machine Learning (ML) and MLOps (Machine Learning Operations).
Duration – 1 day
Delivery – in Classroom, On Site, Remote
PC and SW requirements:
- Internet connection
- Web browser, Google Chrome
- Zoom
Language
- Instructor: English
- Workshops: English
- Slides: English