Senior-Level Production ML Architecture Refresher

Trace the lifecycle of an ML prediction from data ingestion through monitoring in a production architecture.

Distinguish batch, streaming, and online inference by latency, throughput, and consistency requirements.

Select an appropriate consistency model for features, predictions, and model metadata in a given system scenario.

Identify training-serving skew in a production ML design.

Define service-level indicators and objectives for an ML-powered service.

Construct a failure-mode table for the data, model, infrastructure, and application layers.

Estimate the latency, throughput, availability, and cost requirements of an ML workload.