Vertex AI SDK class overview

Data scientists and machine learning (ML) developers use theVertex AI SDK for Python to build, train, and deploy models in a custom MLworkflow. This includes creating datasets and uploading data, training an MLmodel, uploading and storing your model, deploying your model, running batchprediction jobs, and managing your models and endpoints.

The Vertex AI SDK also includes classes to create generative AIsolutions with text, code, chat, and text embedding foundation models. You canuse these classes to generate text, create a text or code chatbot, tune afoundation model, and create a text embedding. A text embedding is text in theform of a vector used to search for items. For more information, seeIntroduction to language model classes in the Vertex AI SDK.

You can use the Vertex AI SDK for Python in hosted JupyterLab notebooks withinVertex AI to write and run your code. The notebooks include preinstalledML frameworks, such as TensorFlow and PyTorch. You can also use other notebooks,such as Colab notebooks, or use a developer environment of your choice thatsupports Python.

If you want to try using the Vertex AI SDK for Python right now, see the followingresources:

The Vertex AI SDK includes many classes to help you automate dataingestion, train models, and get predictions. It also includes classes to helpyou monitor, evaluate, and optimize your machine learning (ML) workflow. Theclasses can be loosely grouped into the following categories:

  • Data classes include classes that work with structured data,unstructured data, and the Vertex AI Feature Store.
  • Training classes include classes that work with AutoMLtraining for structured and unstructured data, custom training,hyperparameter training, and pipeline training.
  • Model classes work with models and model evaluations.
  • Prediction classes work with batch predictions, onlinepredictions, and Vector Search predictions.
  • Tracking classes work with Vertex ML Metadata,Vertex AI Experiments, and Vertex AI TensorBoard.

Except as otherwise noted, the content of this page is licensed under theCreative Commons Attribution 4.0 License, and code samples are licensed under theApache 2.0 License. For details, see theGoogle Developers Site Policies. Java is a registered trademark of Oracle and/or its affiliates.

Last updated 2025-11-24 UTC.