Mlops
MLOps is the discipline of applying engineering and operational practices to develop, deploy, monitor, maintain, and manage machine-learning systems throughout their life cycle.
Machine Learning Platform Engineering
MLOps is the practice of applying software engineering, automation, and operational processes to develop, deploy, and maintain machine-learning systems.
- How it is used
- Frames production ML as an iterative workflow that connects business requirements, data, models, deployment, and monitoring.
- How deep it goes
- Develops the closed-loop workflow and its operational rationale, maps supporting tools, and defers their detailed implementation.
- Examples
- Uses an evolving used-car price predictor and projects involving OCR, movie recommendations, and retrieval-augmented documentation assistance.
- What you can do after
- Apply it via Kubeflow Pipelines, MLflow, and Feast to coordinate pipelines, model management, and reusable features.
Read how Machine Learning Platform Engineering explains mlops →
Designing Deep Learning Systems: A guide for software engineers
A set of practices that bridges machine learning experimentation and production operations to streamline deployment, monitoring, and maintenance.
- How it is used
- Connects experimentation, training, serving, and tracking into reliable delivery of deep learning features.
- How deep it goes
- Defines productionization and enumerates pipelines, reproducibility, metadata traceability, monitoring, feedback, and retraining work.
- Examples
- Uses research-to-product and MLOps-productionization figures plus metadata-reporting and pipeline tasks.
- What you can do after
- Evaluate/understand productionization requirements for a model, from reproducible pipelines through postrelease monitoring and retraining.
Read how Designing Deep Learning Systems: A guide for software engineers explains mlops →
The Art of AI Product Development
Machine learning operations is the set of engineering practices for developing, deploying, monitoring, maintaining, and governing machine-learning systems.
- How it is used
- Accelerates machine-learning iteration by automating monitoring, evaluation, error tracking, and model updates.
- How deep it goes
- Explains lifecycle responsibilities and automation targets, but stops short of detailed implementation procedures or infrastructure designs.
- Examples
- Uses data monitoring, error tracking, evaluation, model updates, and open-source deployment infrastructure as concrete operational examples.
- What you can do after
- Operate machine-learning workflows that integrate new data, evaluate performance, track failures, and deliver model updates more rapidly.
LLMs in Production
Machine learning operations is the discipline of reliably deploying, monitoring, maintaining, and governing machine-learning systems throughout their life cycle.
- How it is used
- Organizes the reliable deployment, maintenance, monitoring, and governance of machine-learning systems across their life cycle.
- How deep it goes
- Explains lifecycle principles and infrastructure layers, then names practices and tools without detailing implementation procedures.
- Examples
- Uses a laptop-based scikit-learn system, Airflow, Prefect, Mage, data pipelines, containers, GPUs, and an LLM infrastructure diagram.
- What you can do after
- Evaluate production needs by separating the machine-learning life cycle from the infrastructure required to support it.
Delivering Machine Learning Projects: From design to deployment
MLOps is the application of development, operations, automation, and lifecycle-management practices to machine learning systems.
- How it is used
- Structures machine-learning delivery by adapting agile and DevOps practices to production-system constraints.
- How deep it goes
- Outlines project setup, data-pipeline concerns, and production design considerations without detailing all four phases.
- Examples
- Uses project setup questions and an ML project-flow diagram to organize goals, constraints, evaluation, data, and dependencies.
- What you can do after
- Evaluate ML projects against production requirements, delivery constraints, data-pipeline risks, privacy, and bias.
Read how Delivering Machine Learning Projects: From design to deployment explains mlops →
Generative AI in Action
MLOps is the set of practices for developing, deploying, monitoring, and maintaining machine-learning models in operational environments.
- How it is used
- Applies DevOps practices to move machine-learning applications from experimentation into reliable, scalable production operation.
- How deep it goes
- Covers lifecycle practices, monitoring, versioning, security, and feedback loops without detailing implementation tools or mathematical model internals.
- Examples
- Uses data, code, models, experiments, inference environments, features, monitoring, and production applications as operational artifacts.
- What you can do after
- Operate machine-learning lifecycles with reproducibility, monitoring, scalability, security, and continuous improvement.
Machine Learning System Design
A set of practices for operationalizing machine-learning systems through reproducibility, testing, automation, deployment, and pipeline management.
- How it is used
- Operationalizes machine-learning systems through training, inference, deployment, and monitoring pipelines.
- How deep it goes
- Explains pipeline-platform requirements and stops short of detailing any specific tool or platform.
- Examples
- Uses Python and shell scripts, MLflow, Kubeflow, BentoML, and cloud ML platforms as implementation examples.
- What you can do after
- Build it from scratch as a scripted training pipeline or configure specialized tools around reproducibility, dependencies, compute, and artifacts.
Related concepts
- LLMOpsincludes · LLMOps is the set of engineering and operational practices for developing, deploying, evaluating, monitoring, and maintaining large language model systems and applications.
- experiment trackinguses · Experiment tracking is the systematic recording and organization of experiment configurations, metrics, artifacts, and results to support inspection, comparison, or reproducibility.
- automationhas part · Automation is the execution of tasks or processes by machines or software with reduced direct human intervention.
- Metaflowincludes · Metaflow is an open-source framework for building, executing, and managing data science and machine learning workflows.
- DevOpscontrasts with · DevOps is a set of organizational, cultural, and technical practices that integrates software development and IT operations through collaboration, automation, and feedback.
- productionizationalternative to · Productionization is the process of preparing, deploying, and maintaining a model or software system for reliable operation in a production environment.
These guides are generated from the text of Manning books by Manning's knowledge graph and link to the sections they draw on. The full books are available on liveBook and at manning.com.