Exam Prep
Google Cloud · PML · Study references

Professional Machine Learning / AI Engineer

The vendor’s own documentation for each objective of the exam, free to use.

Architecting low-code AI solutions2 objectives

Developing ML models using BigQuery ML or AutoML on Gemini Enterprise Agent Platform12 pages

  • BigQuery ML Demand Forecasting

    Shows how to build demand forecasting models with BigQuery ML's ARIMA time-series models using SQL, through a reference pattern on sales data.

  • BigQuery ML Forecasting Overview

    Compares BigQuery ML forecasting models and functions, showing which models support covariates, anomaly detection, training requirements, and evaluation methods.

  • BigQuery ML Inference Models

    Shows how to call Google Vision, Translation, and Natural Language AI services from BigQuery ML using SQL functions for unstructured data analysis.

  • BigQuery ML Introduction

    Understand BigQuery ML, BigQuery's platform for creating, running, and managing generative AI and ML models.

  • BigQuery ML Model Registry Integration

    By registering BigQuery ML models in the Model Registry, you can manage them alongside other ML models, easily version, evaluate, and deploy them for online inference within a unified platform.

  • BigQuery ML Time Series Forecasting

    Shows how to build, evaluate, and forecast with ARIMA time-series models in BigQuery ML, covering model options, evaluation metrics, and coefficient inspection.

  • BigQuery overview and features

    Introduces BigQuery as a serverless analytics platform, outlining its architecture, data loading and querying capabilities, SQL support, and available interfaces and tools.

  • BigQuery Scheduled Queries

    Explains how to create, schedule, and manage BigQuery scheduled queries, including runtime parameters, required permissions, and destination table behavior.

  • Continuous Model Evaluation Setup

    Shows how to set up continuous model evaluation by combining BigQuery ML stored procedures, Cloud Functions, and Cloud Scheduler to retrain and score models on a schedule.

  • ML.FORECAST function

    Describes the ML.FORECAST function syntax and its parameters for generating time-series forecasts from trained ARIMA_PLUS models and their XREG variant.

  • Training and Using ML Models

    Develop and deploy ML models on Agent Platform. Choose AutoML, run custom training with serverless jobs or dedicated clusters, or scale with Ray.

  • Training Methods Comparison

    Learn about the key differences between training a model in Gemini Enterprise Agent Platform using AutoML, custom training, Ray on Gemini Enterprise Agent Platform, or training a model using BigQuery ML.

Building AI solutions using Google Cloud AI APIs or foundational models4 pages

  • Generative AI Documentation Overview

    Documentation and resources for building and implementing generative AI applications with Google Cloud tools and productss.

  • Google Cloud AI APIs

    Browse Google Cloud's AI APIs for everything from generative AI to machine learning to conversational AI.

  • Google Cloud AI Products

    Easy-to-use scalable AI offerings including Gemini platform, video and image analysis, speech recognition, and vision AI.

  • Model Garden Overview

    Discover, customize, and deploy AI models from Google and partners using Gemini Enterprise Agent Platform (formerly Vertex AI).

Collaborating within and across teams to manage data and models3 objectives

Exploring and preprocessing data for ML4 pages

  • BigQuery ML Feature Engineering

    How to preprocess data using BigQuery ML so you can get better insights and models.

  • Dataflow ML Notebooks Resources

    Explore Dataflow ML notebooks for Apache Beam pipelines. Discover tutorials for machine learning workflows and data processing.

  • MLTransform Embedding Generation

    Describes how to use MLTransform to generate text embeddings and chunk documents in Dataflow pipelines, with example code and vector database loading.

  • Plan Dataflow Pipeline

    Explains how to plan Dataflow pipelines around service level objectives, isolation of data and resources, external service calls, and Pub/Sub usage.

Model prototyping using notebooks (e.g., Gemini Enterprise Agent Platform Workbench and Colab Enterprise)6 pages

Tracking and running ML experiments4 pages

Scaling prototypes into ML models3 objectives

Building models given the task considering cost, complexity, latency, and scalability2 pages

Training models3 pages

  • AI Model Concepts and Lifecycle

    Introduces AI models, their categories, and the lifecycle stages from training through evaluation, useful for grasping foundational AI terminology and concepts.

  • AI Model Fundamentals

    AI models are programs trained to recognize patterns and make decisions. Learn how AI models work and how to deploy your own with Google Cloud.

  • Training and Using ML Models

    Develop and deploy ML models on Agent Platform. Choose AutoML, run custom training with serverless jobs or dedicated clusters, or scale with Ray.

Choosing appropriate hardware for training9 pages

  • Accelerator-optimized GPU machine families

    Understand accelerator-optimized machine families for GPU-accelerated AI, ML, and HPC workloads.

  • AI Hypercomputer GPU Infrastructure Categories

    Identify the optimal accelerator infrastructure for AI and ML workloads.

  • AI/ML Performance Optimization

    Provides performance optimization principles and recommendations for AI and ML workloads.

  • GPU Memory Consumption Facts

    Explains GPU high-bandwidth memory consumption when fine-tuning AI models, covering weights, optimizer states, gradients, activations, and how LoRA and quantization reduce usage.

  • GPU Models and Specifications

    Learn about GPUs on Google Cloud, including GPU acceleration fundamentals, available GPU models, machine series mapping, performance characteristics, and pricing considerations.

  • Multi-instance GPUs on GKE

    Explains how to partition NVIDIA GPUs into multi-instance GPU slices on GKE, covering supported GPU types, version requirements, and partition syntax.

  • Production ML Pipeline Preparation

    Outlines preparing production ML pipelines, covering hardware and quota planning, and what each pipeline should monitor for skew, drift, and outages.

  • PyTorch Training on Vertex AI

    Compares DataParallel, DistributedDataParallel, and FullyShardedDataParallel for PyTorch training on Vertex AI, guiding readers on choosing a multi-GPU strategy.

  • Reduction Server PyTorch Training

    Explains how Reduction Server speeds up multi-node distributed GPU training with PyTorch, describing its reducer role and compatibility with DDP and FSDP.

Serving and scaling models2 objectives

Serving models7 pages

  • AI Inference Model Serving Networking

    Describes reference networking architecture for serving AI inference on GKE, covering private endpoints, the inference gateway, API management, and content screening.

  • AI Inference Unified Frontend Architecture

    Describes a unified frontend architecture for serving AI models through a single private endpoint, showing how requests are routed to model backends by model name.

  • AI Platform Custom Online Prediction

    Walks through building a custom online prediction routine for a deployed text classification model, covering preprocessing, model setup, and serving code structure.

  • MLOps Definition and Components

    Defines MLOps and its core components, contrasting it with DevOps, and outlines the practices for managing the machine learning lifecycle.

  • PyTorch Vertex AI Deployment

    Explains how to deploy PyTorch models on Vertex AI Prediction using TorchServe, covering model archiving, custom handlers, and container requirements.

  • Serverless ML Model Deployment

    Walks through deploying a trained ML model as a serverless endpoint using Cloud Functions, covering API client setup, request payloads, and model metadata lookups.

  • Vertex AI Feature Store Online Serving

    Compares Vertex AI Feature Store online serving options, covering data volume suitability, sync modes, embeddings support, and settings needed to serve null feature values.

Scaling online model serving2 pages

Automating and orchestrating ML pipelines2 objectives

Developing end-to-end ML pipelines5 pages

Automating model retraining17 pages

  • BigQuery ML Monitoring Functions

    Describes BigQuery ML functions for describing tables and detecting data skew and drift, guiding readers on monitoring model input data quality over time.

  • Continuous Model Evaluation Setup

    Shows how to set up continuous model evaluation by combining BigQuery ML stored procedures, Cloud Functions, and Cloud Scheduler to retrain and score models on a schedule.

  • Data Drift Detection Pipeline

    Describes an event-triggered pipeline that uses TensorFlow Data Validation and Kubeflow Pipelines to detect data drift when new files arrive in Cloud Storage.

  • Feature Attribution Monitoring

    Explains feature attribution monitoring for deployed models, covering how attributions are computed, chosen for classes, and configured with gcloud monitoring jobs.

  • Federated Learning Architectures on Google Cloud

    Explains federated learning architectures on Google Cloud, comparing cross-silo and cross-device designs and the horizontal, vertical, and transfer learning variants.

  • Google Cloud Labelbox Architecture

    Shows a reference architecture integrating Labelbox data labeling with Cloud Storage, BigQuery, and Vertex AI for building and managing training datasets.

  • ML Implementation Best Practices on Google Cloud

    Introduces best practices for implementing machine learning (ML) on Google Cloud, with a focus on custom-trained models based on your data and code.

  • ML Pipelines Production Architecture

    Describes the production ML pipeline architecture for serving, data, training, and validation, explaining how they connect for reliable predictions over time.

  • ML Solution Quality Guidelines

    Collates some guidelines to help you assess, ensure, and control quality in machine learning (ML) solutions.

  • MLOps Architecture with TFX and Pipelines

    Describes the architecture of a machine learning (ML) system using TensorFlow Extended (TFX) libraries.

  • MLOps Definition and Components

    MLOps, or machine learning operations, aims to deploy and maintain machine learning models reliably and efficiently. Read on to learn more.

  • MLOps: CI/CD and CT for ML

    Explains MLOps practices for continuous integration, delivery, and training of ML systems, and compares manual level 0 with automated level 1 pipeline maturity.

  • Model Monitoring Overview

    Overviews Model Monitoring v1 and v2 for Agent Platform, comparing monitoring objectives, supported metrics, prerequisites, and migration considerations.

  • Model Monitoring Setup and Configuration

    Explains how to set up and configure model monitoring jobs, covering CLI commands, default thresholds and intervals, BigQuery logging, and Cloud Logging for alerts.

  • Triggering Vertex Pipeline Runs

    Shows how to compile a Vertex AI pipeline and trigger its runs automatically with a Cloud Function and Cloud Scheduler based on new BigQuery data.

  • Vertex AI Forecasting Updates

    Announces new Vertex AI forecasting features, including the TiDE model, larger dataset support, pipeline-based backend, and probabilistic inference, for readers evaluating upgrades.

  • Vertex AI Model Monitoring

    Introduces Vertex AI Model Monitoring concepts and setup, explaining how to configure drift detection jobs, data sources, and alert notifications for deployed models.

Monitoring AI solutions2 objectives

Identifying risks to AI solutions5 pages

  • AI Protection overview

    Get an overview of AI Protection, its benefits, and how it helps protect AI workloads.

  • AI Risk and Resilience 2025

    Explore the previous Mandiant AI risk and resilience report covering 2025 AI threat trends, shadow AI governance, and SOC defense.

  • AI Risk and Resilience 2026

    Read our 2026 AI risk and resilience report to explore adversarial AI tactics, red teaming insights, and modern defense architecture.

  • AI-Assisted Vulnerability Management

    Details the architectural risks of integrating LLM agents into CI/CD pipelines for automated vulnerability discovery.

  • AI/ML Security Best Practices

    Security principles and recommendations for the AI and ML perspective in the Well-Architected Framework.

Monitoring, testing, and troubleshooting AI solutions4 pages

  • AI/ML Performance Optimization

    Provides performance optimization principles and recommendations for AI and ML workloads.

  • Database Operations Agents

    A closer look at AI-powered database operations agents for setting up and onboarding AlloyDB, Bigtable, Cloud SQL, Spanner, as well as managing them.

  • Gemini Cloud Assist Features

    AI assistance for managing applications across the lifecycle - design, deployment, monitoring, troubleshooting, performance and cost optimization.

  • Model Monitoring Metrics

    Lists model observability metrics in Metrics Explorer and explains what the dashboard collects, its limitations, and how to filter by model endpoint.

All certifications

Terms of use

  • This list is free to use. Every link goes to the vendor’s own site; we don’t control those pages and they may change or move.
  • Each page’s summary is ours, written to say what the page covers. It isn’t the vendor’s and may be imperfect. The list is provided as is, without warranty, and isn’t endorsed by any vendor.
  • We count visits and the links opened, against a random id kept in your browser. Nothing personal is recorded. If you sign up, that id is tied to your account so we can tell whether this list is useful.
  • Anything you submit to this list may be published.