Feature stores, streaming versus batch, point-in-time correctness and lineage.
8h · 7 lessons · 1 challenge
Before this: Data Pipelines & Feature Engineering
The infrastructure that makes the same feature available offline for training and online for serving, computed identically, at the right point in time. Get it wrong and every model on top of it is quietly wrong.
Covers the build-versus-buy question honestly — most teams do not need a feature store, and knowing when you do is the actual skill.
Reconstruct a training set exactly as it was on a past date, with no future information leaking in. The grader checks every row.
Worth 550 XP.
