Practical MLOps: From Notebook to Production
Packaging, deployment, monitoring and retraining that lasts
By Houssam Kodad
One-time purchase
€32.95
VAT included
where applicable
- Instant download after purchase
- Readable on any device
- Free updates to this edition
- Secure checkout
About this book
What's inside
A model that wins offline is worth nothing until it serves reliably and keeps working. This book is an end-to-end MLOps playbook for engineers: reproducible training, versioned data and models, automated deployment, and the monitoring that catches drift before your users do. It treats the model as one component in a system that has to be operated, not a science project that ends at the leaderboard.
What you'll learn
Skills you'll walk away with
- Make training reproducible with versioned data and code
- Track experiments, metrics and model lineage
- Package models for batch and online serving
- Automate deployment with CI/CD for ML
- Monitor data drift, concept drift and performance decay
- Design shadow, canary and A/B rollout strategies
- Build retraining pipelines with human-in-the-loop gates
- Set up governance, rollback and incident response
Table of contents
10 chapters-
01
What MLOps Is Really Solving
- · The maintenance cost of models
- · Maturity levels
- · A system view of ML
-
02
Reproducible Training
- · Versioning data and features
- · Deterministic pipelines
- · Environment and dependency pinning
-
03
Experiment Tracking and the Model Registry
- · Logging runs and metrics
- · Comparing and promoting models
- · Model lineage and approval
-
04
Packaging Models for Serving
- · Batch vs online inference
- · Containerising the model
- · Contracts for inputs and outputs
-
05
CI/CD for Machine Learning
- · Testing models and pipelines
- · Automated build and deploy
- · Promotion across environments
-
06
Serving at Low Latency
- · Real-time inference services
- · Feature lookups at request time
- · Scaling and caching
-
07
Monitoring Models in the Wild
- · Data and prediction drift
- · Performance decay detection
- · Ground-truth delay
-
08
Safe Rollouts
- · Shadow deployments
- · Canary and A/B testing
- · Automatic rollback
-
09
Retraining Without Regret
- · Triggers for retraining
- · Validation gates
- · Champion/challenger evaluation
-
10
Governance and Incident Response
- · Audit trails and approvals
- · On-call for models
- · A production readiness checklist
This is the full chapter list — exactly what you'll receive in the PDF.
More in Data Science & ML
Keep exploring this track
Feature Engineering for Machine Learning
Crafting, selecting and serving features that move metrics
Time Series Forecasting in Practice
Classical models, gradient boosting and deep learning for demand and operations
Recommender Systems at Scale
Collaborative filtering, embeddings and multi-stage ranking
Statistical Foundations for Data Scientists
Inference, experimentation and A/B testing done right
Gradient Boosting with XGBoost and LightGBM
A practitioner's guide to winning with tabular data