Domain: MLOps & Infrastructure

15 terms

Batch Inference Intermediate

Running predictions on large datasets periodically.
Blue-Green Deployment Intermediate

Maintaining two environments for instant rollback.
Canary Release Intermediate

Incrementally deploying new models to reduce risk.
CI/CD for ML Intermediate

Automated testing and deployment processes for models and data workflows, extending DevOps to ML artifacts.
Data Drift Intermediate

Shift in feature distribution over time.
Domain Shift Intermediate

A mismatch between training and deployment data distributions that can degrade model performance.
Feature Store Intermediate

Centralized repository for curated features.
Feedback Loop Intermediate

Using production outcomes to improve models.
Inference Pipeline Intermediate

Model execution path in production.
MLOps Intermediate

Practices for operationalizing ML: versioning, CI/CD, monitoring, retraining, and reliable production management.
Monitoring Intermediate

Observing model inputs/outputs, latency, cost, and quality over time to catch regressions and drift.
Online Inference Intermediate

Low-latency prediction per request.
Prediction Drift Intermediate

Shift in model outputs.
Shadow Deployment Intermediate

Running new model alongside production without user impact.
Training Pipeline Intermediate

End-to-end process for model training.