MLOps on Kubernetes Training in Madrid, Spain

The MLOps on Kubernetes Training is a hands-on, enterprise-grade program focused on building, deploying, scaling, and operating machine learning systems on Kubernetes.
Participants will learn how to operationalize ML models using containerization, CI/CD, model lifecycle management, monitoring, and scalable inference/training, enabling reliable, repeatable, and production-ready ML platforms.

Introduction to MLOps & Kubernetes

• What is MLOps and why it matters

• Challenges of production ML systems

• Why Kubernetes for ML platforms

• MLOps lifecycle overview

Kubernetes Fundamentals for MLOps

• Kubernetes architecture recap

• Pods, Deployments, Services for ML workloads

• Namespaces and resource isolation

• Storage and networking basics for ML

Containerizing ML Workloads

• Packaging training and inference code

• Building ML-ready Docker images

• Managing dependencies and environments

• Image versioning and reproducibility

ML Pipelines on Kubernetes

• Pipeline concepts and orchestration

• Kubeflow Pipelines overview

• DAG-based ML workflows

• Managing pipeline artifacts and metadat

Experiment Tracking & Model Registry

• MLflow tracking and experiments

• Model versioning and lineage

• Model registry concepts

• Promoting models across environments

Training at Scale on Kubernetes

• Distributed training concepts

• GPU and accelerator scheduling

• Resource requests and limits

• Managing large-scale training job

Model Serving & Inference

• Batch vs real-time inference

• Deploying inference services

• Canary and blue/green model deployments

• Autoscaling inference workloads

CI/CD for Machine Learning

• CI/CD pipelines for ML code and models

• Automating training and deployment

• GitOps for ML workflows

• Environment promotion strategies

Data & Feature Management

• Data versioning concepts

• Feature stores overview

• Managing training vs inference data

• Data validation and quality checks

Monitoring, Drift & Observability

• Monitoring model performance

• Data and concept drift detection

• Logging and metrics for ML systems

• Alerting and incident response

Security, Governance & Compliance

• Secrets and credential management

• Access control for ML platforms

• Model auditability and traceability

• Compliance and responsible AI considerations

Cost, Performance & Optimization

• Resource optimization strategies

• GPU and compute cost management

• Caching and reuse of artifacts

• Capacity planning for ML workloads

Cloud & Hybrid MLOps Patterns

• Running MLOps on cloud Kubernetes

• Hybrid and multi-cluster strategies

• Centralized vs decentralized ML platforms

• Integration with cloud ML services

Hands-on Exercises

Summary and Conclusion