Batch vs Online Inference

Day 56 of 100 Days of MLOps. Your API serves one prediction per request — but sometimes you need to score millions of rows at once, and doing that one HTTP call at a time would be absurd. Learn the two serving modes: online (single request, low latency) and batch (a whole dataset scored at once, hugely efficient). Build a batch scoring job from the same model and see it run thousands of times faster per row. Runs 100% locally on any OS.