Compare online serving, batch serving, and streaming serving for ML models. When would you use each approach?

Sample answer preview

Model serving patterns determine how predictions reach end users and applications. The choice between online, batch, and streaming serving depends on latency requirements, prediction volume, and how predictions integrate into business workflows.