Agent Platform for Machine Learning Practitioners
This 1-day course is a practical, hands-on introduction to Agent Platform for machine learning practitioners who want to run custom model workflows.
Agent Platform for Machine Learning Practitioners is designed for engineers and data scientists who already know ML models and want to become proficient with Agent Platform. You learn how its key components fit together to support your ML workflows, then work through containerized training applications, custom training and hyperparameter tuning jobs, versioning your models in the Model Registry and deploying them for online predictions with Agent Platform Endpoints.
The course then covers orchestrating end-to-end workflows with Agent Platform Pipelines, including Kubeflow, pre-built and lightweight Python components, and compiling and running pipelines. It closes with model monitoring for deployed models, including feature drift and skew. The day combines three lectures with five hands-on labs.
The course is aimed at machine learning engineers and data scientists. Participants should have experience building and training custom ML models and be familiar with Docker.