ML Feature Stores in Regulated Environments: Lineage, Drift, and the Model Risk Problem

SR 11-7

Federal Reserve model risk management guidance requiring documented input lineage for all regulated models

Feature stores â€” Feast, Hopsworks, Tecton â€” solve the operational ML problem of feature consistency between training and serving. In regulated environments, they solve additional problems: feature lineage documentation for SR 11-7 model risk management, feature drift detection as an ongoing monitoring requirement under FDIC model risk guidance, and point-in-time correct feature retrieval for backtesting regulated models without data leakage. The GDPR problem is less obvious: PII features used in regulated models must have documented lawful basis, retention limits, and deletion capability â€” which is architecturally difficult when features are precomputed and cached.

Full article content coming soon.

Compliance Engineering

The engineering behind this article is available as a service.

We have done this work — not advised on it, not reviewed documentation about it. If the problem in this article is your problem, the first call is with a senior engineer who has solved it.

Talk to an Engineer See Case Studies →

Related Reading

ML Feature Stores in Regulated Environments: Lineage, Drift, and the Model Risk Problem

EU AI Act: What CTOs Actually Need to Do Before August 2026

What Happens to Your HIPAA BAAs When You Migrate to Cloud

The Vendor Rescue Pattern: How to Recover a Failed Implementation in 12 Weeks

The engineering behind this article is available as a service.