Google Cloud
Google Cloud ML & cert prep — the Professional Machine Learning Engineer (PMLE) learning path: designing, deploying, and monitoring ML on Vertex AI. Blueprint, roadmap, and free-first resources are live; decks and a practice exam are in build.
Flashcard decks
8 decks · 165 cards · in study orderGCP infrastructure & interview drill
2 decks · 66 cards1GCP for Cloud Infrastructure Engineers (projects, IAM, VPC networking, GKE, ops)Interview-ready cards on Google Cloud Platform for a Cloud Infrastructure Engineer coming from Azure/AWS: the resource hierarchy (org/folders/projects), Cloud IAM + service accounts + Workload Identity, GCP's global VPC networking model, Compute Engine/GKE/Cloud Run, load balancing, security controls (VPC Service Controls, Org Policy, Assured Workloads), IaC with Terraform, and the Cloud Operations suite. Each card translates the GCP term to its Azure/AWS equivalent where it helps.342GCP Cloud Infrastructure — Interview DrillRapid-recall drill for a GCP cloud-infrastructure interview when your production depth is Azure/AWS: the service Rosetta table, the structural differences that actually trip people up (global VPC, project boundary, service-account identity), federal compliance on GCP, and the triage answers. Built for the night before.32
PMLE exam sections
6 decks · 99 cards3Google PMLE · Scaling Prototypes into ML Models (~21%)The biggest section: turning a notebook prototype into a real, scalable Vertex AI model — framing the problem, choosing a build approach (custom training, AutoML, BigQuery ML, prebuilt/foundation), the right framework & hardware (CPU/GPU/TPU), distributed & efficient training, hyperparameter tuning, and evaluating against the business goal.214Google PMLE · Serving & Scaling Models (~20%)Getting models to production and keeping them fast and cheap: online vs batch prediction on Vertex AI, endpoints & deployed models, autoscaling and machine/accelerator sizing, traffic splitting for safe rollout and A/B, private endpoints, optimizing latency/throughput/cost, and serving generative-AI models.195Google PMLE · Automating & Orchestrating ML Pipelines (~18%)MLOps automation on GCP: Vertex AI Pipelines (KFP/TFX), components & artifacts, CI/CD for ML, triggers & scheduling, retraining loops, the Model Registry + evaluation gates, and the orchestration glue (Cloud Build, Artifact Registry, Cloud Scheduler, Pub/Sub, Eventarc) that turns manual training into a repeatable system.186Google PMLE · Collaborating to Manage Data & Models (~15%)The data & governance layer for team ML: Vertex AI Feature Store (consistent features, no training/serving skew), managed Datasets, data governance & lineage (Dataplex, BigQuery), IAM & least-privilege, the Model Registry & metadata, responsible data handling (PII, VPC-SC), and reproducible team workflows.157Google PMLE · Architecting Low-Code AI Solutions (~13%)The fast paths to an AI solution without heavy custom code: BigQuery ML (train with SQL), Vertex AI AutoML, pre-trained/foundation APIs (Vision, Natural Language, Speech, Translation, Document AI), Gemini & Model Garden, and Vertex AI Search & agent-building — knowing which low-code option fits which problem.138Google PMLE · Monitoring AI Solutions (~13%)Keeping production ML healthy and responsible: Vertex AI Model Monitoring for feature drift and training/serving skew, prediction/endpoint observability with Cloud Monitoring & Logging, retraining triggers, explainability (Vertex Explainable AI), responsible-AI & safety for GenAI, and cost/performance monitoring.13