Training-Serving Skew
A discrepancy between how features are computed during model training versus how they are computed during production serving. This is one of the most common and hardest-to-detect causes of model failure.
Why It Matters
Training-serving skew silently degrades model performance. Models appear to work in testing but fail in production because the data pipeline differs.
Example
A model trained with features computed in batch (using future data that would not be available in real time) that then receives different feature values in real-time serving.
Think of it like...
Like practicing basketball with a regulation ball but playing the game with a slightly different ball — the differences are subtle but affect performance.
Related Terms
Feature Store
A centralized repository for storing, managing, and serving machine learning features. It ensures consistent feature computation between training and serving, and enables feature reuse across teams.
MLOps
Machine Learning Operations — the set of practices that combine ML, DevOps, and data engineering to deploy and maintain ML models in production reliably and efficiently.
Model Monitoring
The practice of continuously tracking an ML model's performance, predictions, and input data in production to detect degradation, drift, or anomalies after deployment.
Data Pipeline
An automated workflow that extracts data from sources, transforms it through processing steps, and loads it into a destination for use. In ML, data pipelines ensure consistent data flow from raw sources to model training.
Deployment
The process of making a trained ML model available for use in production applications. Deployment involves packaging the model, setting up serving infrastructure, and establishing monitoring.