PMLE study notes
- Architecting low-code AI solutions PMLE notes on BigQuery ML model types, feature engineering and prediction, ML APIs and Model Garden, Document AI and industry APIs, RAG with Agent Builder, and AutoML for tabular, image, text, video and forecasting. 13%
- Collaborating to manage data and models PMLE notes on organising and preprocessing data, Vertex AI datasets, Dataflow and TFX, Feature Store, PII and PHI handling, notebook backends, Workbench security, Model Garden and experiment tracking. 14%
- Scaling prototypes into ML models PMLE notes on choosing frameworks and architectures, interpretability, training data and ingestion, Vertex AI custom training, distributed training, hyperparameter tuning, troubleshooting, fine-tuning foundation models and choosing CPUs, GPUs and TPUs. 18%
- Serving and scaling models PMLE notes on batch and online inference, Vertex AI endpoints, model registry, A/B testing and traffic splits, prediction containers, Feature Store lookups, autoscaling, hardware, and optimising models for latency, memory and throughput. 20%
- Automating and orchestrating ML pipelines PMLE notes on end-to-end ML pipelines, data and model validation, train-serve consistency, Vertex AI Pipelines, Kubeflow, TFX, Cloud Composer, MLflow, triggers, retraining policy, CI/CD for ML and metadata and lineage. 22%
- Monitoring AI solutions PMLE notes on securing AI systems, Responsible AI and fairness, explainability on Vertex AI, Model Monitoring for skew and drift, feature attribution drift, performance against baselines, and troubleshooting training and serving errors. 13%