📄️ GCP Vertex AI
Train, tune, register, serve and orchestrate machine learning models on GCP Vertex AI, from BigQuery data to a KFP pipeline in production, with a 40-question exam and a verifiable certificate.
📄️ 1. Components and vocabulary
Module 1 of the GCP Vertex AI premium course: projects, regions, APIs to enable, service accounts, the map of Vertex services and per-second billing you can actually read.
📄️ 2. Managed notebooks
Module 2 of the GCP Vertex AI premium course: Workbench instances, idle shutdown, kernels, direct BigQuery access from the notebook, and Colab Enterprise as an alternative.
📄️ 3. Cloud Storage and BigQuery
Module 3 of the GCP Vertex AI premium course: bucket layout, BigQuery table design, managed datasets, DataFrame export for training and the true dollar cost of a query.
📄️ 4. Custom training
Module 4 of the GCP Vertex AI premium course: prebuilt vs custom containers, CustomTrainingJob, machine types and accelerators, and where the trained model artifact lands on GCS.
📄️ 5. Hyperparameter tuning
Module 5 of the GCP Vertex AI premium course: Vizier's Bayesian search, metrics reported by the code, parallel trials, early stopping and how to read the trial results.
📄️ 6. Model Registry
Module 6 of the GCP Vertex AI premium course: uploading a model, versions, the default alias, attached evaluation and lineage back to the training job and dataset.
📄️ 7. Endpoints and scaling
Module 7 of the GCP Vertex AI premium course: deploying two model versions on one endpoint with 90/10 split, min and max replicas, latency and request logging.
📄️ 8. Batch prediction
Module 8 of the GCP Vertex AI premium course: BigQuery-to-BigQuery batch prediction, scheduling, cost compared to online serving and joining scores back to the source table.
📄️ 9. Vertex AI Pipelines
Module 9 of the GCP Vertex AI premium course: KFP components, compilation, execution on Vertex, artifacts and lineage, a conditional deployment gate, and scheduling.
📄️ 10. Foundation models
Module 10 of the GCP Vertex AI premium course: Model Garden, calling a foundation model, managed tuning, classifying dispute comments and reasoning about cost per token.
📄️ Recap and exam
Recap of the GCP Vertex AI premium course: components, notebooks, data, training, tuning, registry, endpoints, batch, pipelines, foundation models — then the 40-question exam.