GoogleCloudPlatform/vertex-ai-samples

Notebooks, code samples, sample apps, and other resources that demonstrate how to use, develop and manage machine learning and generative AI workflows using Google Cloud Vertex AI.

What it solves

This repository provides a comprehensive set of practical examples and resources to help developers and ML practitioners implement machine learning and generative AI workflows on the Google Cloud Vertex AI platform. It bridges the gap between documentation and production-ready implementation by providing runnable notebooks and sample apps.

How it works

The project is organized as a collection of notebooks, code samples, and "skills" that demonstrate specific Vertex AI capabilities. Users can run these examples directly in Colab, Colab Enterprise, or Vertex AI Workbench. The repository is structured to cover the entire ML lifecycle, including data management (Feature Store, BigQuery), model development (AutoML, custom models, Ray on Vertex AI), deployment (Prediction, Model Registry), and operational tools (Pipelines, ML Metadata, Explainable AI).

Who it’s for

  • ML Practitioners: Those looking to build, train, and deploy custom ML models using Google Cloud's infrastructure.
  • Generative AI Developers: Users wanting to implement Gemini, Gemma, and other open-source models from the Model Garden.
  • Cloud Architects: Professionals needing to deploy scalable AI agent "skills" and managed ML pipelines.

Highlights

  • End-to-End Workflow Examples: Covers everything from data labeling and feature stores to model registration and prediction.
  • AI Agent Skills: A dedicated suite of skills for routing tasks, inferencing, and fine-tuning Gemini and open models.
  • Model Garden Integration: Examples for using a wide variety of first-party and third-party models (e.g., Llama 3, Claude 3).
  • Managed Infrastructure: Integration with Colab Enterprise and Vertex AI Workbench for immediate execution.