statmike/vertex-ai-mlops

Google Cloud Platform Vertex AI end-to-end workflows for machine learning operations

What it solves

This project provides a comprehensive library of hands-on workflows for implementing machine learning operations (MLOps) and generative AI on Google Cloud Platform (GCP). It bridges the gap between theoretical documentation and production-hardened code by providing adaptable, interactive notebooks that demonstrate how to integrate Vertex AI with the broader GCP ecosystem (such as BigQuery, Dataflow, and various database services).

How it works

The repository is organized into a massive collection of over 470 interactive notebooks categorized by functional area:

  • MLOps: Covers the full lifecycle from training to production, including serving (online/batch), feature stores, KFP pipelines, model evaluation, and monitoring.
  • Data + AI: Demonstrates bringing inference to the data using BigQuery AI functions, Dataflow, and Dataproc.
  • Applied GenAI: Focuses on grounding LLMs through RAG (Retrieval-Augmented Generation), vector search across 11 different GCP databases, embedding strategies, and evaluation metrics.
  • Applied Forecasting: Provides a learning path for time series forecasting using everything from BQML to foundation models like TimesFM.
  • Framework Workflows: Implements end-to-end pipelines for PyTorch, Keras, CatBoost, and R.
  • Applied ML: Features prototypes for AI agents (using ADK and MCP), document processing, and black-box optimization via Vertex AI Vizier.

Who it’s for

  • ML engineers and data scientists working within the Google Cloud ecosystem.
  • Developers looking to build and deploy AI agents or RAG pipelines on Vertex AI.
  • Teams needing practical patterns for scaling ML inference and managing feature stores on GCP.

Highlights

  • Massive Scale: Over 470 interactive notebooks covering a vast array of GCP AI services.
  • Broad Integration: Shows how to connect Vertex AI with BigQuery, Spanner, AlloyDB, and GKE.
  • RAG Deep Dives: Detailed comparisons of vector search across 11 different Google Cloud databases.
  • Agentic Workflows: Practical implementations of AI agents using the Agent Engine and ADK.