Features must be defined once and reused consistently across several models. What do you use?
Why A is the answer
Feature Store on Agent Platform
Feature Store manages and serves shared features, which avoids rebuilding the same logic per model.
In detail
- Feature Store defines features once for reuse.
- Consistent features also reduce skew.
Use case
An illustrative scenario, not a real incident.
Three models need the same 30-day spend figure. Feature Store defines it once and serves it to all three.
See it in different ways
flowchart LR A["Feature pipeline"] --> B["Agent Platform Feature Store"] B --> C["Training"] B --> D["Online serving"]
mindmap
root((ML data))
Processing
Dataflow
Features
Feature Store
Prototyping
Workbench
Colab Enterprise
Tracking
Experiments
ML Metadata lineagemindmap
root(("Feature Store"))
n1["Consumers"]
n2["Churn model"]
n3["Fraud model"]
n4["Pricing model"]Key point: Define features once, and reuse them.
Why the other options are wrong
- B. A Cloud Run service that recomputes features
- Recomputing per service risks skew.
- C. Model Armor with shared feature rules
- Model Armor screens prompts.
- D. A shared CSV file in Cloud Storage
- A CSV has no definitions or serving.