Chapter 12: Deployment & MLOps for Scientific Applications

Introduction: From Notebooks to Production Science

Part I: Experiment Tracking & Management

The Problem: Experiment Chaos

MLflow for Scientific Experiments

Part II: Data Versioning & Lineage

DVC for Scientific Data

Part III: Model Lifecycle Management

Model Registry for Science

Part IV: Continuous Training Pipelines

Automated Retraining with New Data

Part V: Scientific Validation & Testing

Domain-Specific Test Suite

Part VI: Deployment to Scientific Infrastructure

Integration with HPC Clusters

Part VII: Monitoring Production Models

Scientific Drift Detection

References