Model Online / On-Site
Duration
Modules 5 modules

Course Objective & Scope

Routing 100% of live traffic directly to a newly trained model risks schema incompatibilities, sudden latency spikes or silent accuracy regressions that disrupt business operations. This course covers end-to-end model artefact versioning, lineage tracking, progressive rollout patterns and instant rollback mechanisms in production.

Participants manage model aliases (@champion, @challenger) in MLflow Model Registry, configure canary, shadow (dark launch) and A/B traffic routing with KServe/Istio, and trigger automated rollbacks on metric violations. Teams leave with a formal model promotion runbook and rollout checklist for their organisation.

Who Should Attend?

  • MLOps engineers managing model release pipelines and CI/CD/CT workflows
  • ML engineers and data scientists regularly retraining and replacing live models
  • Platform and SRE teams responsible for service reliability and traffic routing
  • Prerequisite: basic knowledge of model deployment, Git and containers/Kubernetes

Key Takeaways

  • Version datasets, training code, hyperparameters and model weights with full lineage
  • Manage model signatures and @champion/@challenger aliases in MLflow Model Registry
  • Apply shadow, canary and blue/green deployment strategies to the right release scenarios
  • Validate business impact using Champion-Challenger setups and online A/B experiments
  • Trigger automated rollbacks when latency, error rate or prediction drift thresholds are breached

Curriculum

Model Versioning, Lineage and the Model Registry
  • Linking code (Git), data (DVC/LakeFS) and model artefacts for full lineage
  • MLflow Model Registry: model signatures, metadata, tags and @champion / @challenger aliases
Schema Evolution and Backward Compatibility
  • Preserving API contracts when features are added or model outputs change
  • Model wrappers, fallback defaults and automated contract tests
Progressive Rollout Patterns: Blue/Green, Canary and Shadow
  • Instant environment switches with Blue/Green and weighted canary splitting via KServe/Istio
  • Shadow (traffic mirroring) deployments to test new models on live requests with zero user impact
Champion-Challenger, A/B Testing and Multi-Armed Bandits
  • Distinguishing technical validation (canary) from business metric validation (A/B testing)
  • Sticky routing, prediction logging and statistical significance evaluation
Automated Rollbacks and CI/CD Model Promotion Pipelines
  • Automating model promotion with Registry webhooks and GitOps (Argo Rollouts)
  • Single-command and automated rollbacks triggered by SLO breaches or prediction drift

Other Courses in This Category

  • Scaling Deployed Machine Learning and LLM Services

    Horizontal and vertical scaling for live model services: autoscaling on queue/GPU metrics with Kubernetes HPA, KEDA and Knative, reducing cold starts, MIG/time-slicing and load balancing. For MLOps, SRE and platform engineers.

    Model Serving & Deployment Corporate
    Live Course
    5 modules
    View
  • Introduction to Model Deployment

    The core steps for moving a trained model from notebook to production: serialization, serving via an API, container packaging, basic monitoring and rollback. For data scientists and developers new to model deployment.

    Model Serving & Deployment Corporate
    Live Course
    5 modules
    View
  • REST API and gRPC Fundamentals for Model Serving

    Fundamentals of REST and gRPC for model services: HTTP/JSON design, OpenAPI, Protocol Buffers, streaming RPCs and performance comparison. For developers who want to pick the right protocol and build their first service.

    Model Serving & Deployment Corporate
    Live Course
    5 modules
    View
  • Synchronous vs Asynchronous Inference

    Synchronous and asynchronous inference patterns: request-response, queue-based processing (Celery, Redis, Kafka), job polling, webhooks and async Python. For engineers designing long-running model calls that scale.

    Model Serving & Deployment Corporate
    Live Course
    5 modules
    View
  • Batch, Real-Time and Streaming Inference Fundamentals

    A comparison of the three core inference modes: batch scoring with Spark and Airflow, REST-based real-time services and streaming inference with Kafka/Flink. For data and ML teams choosing the right mode for a business need.

    Model Serving & Deployment Corporate
    Live Course
    5 modules
    View
  • Model Serialization and Model File Formats

    Saving models safely and portably: pickle/joblib risks, TorchScript, SavedModel, ONNX, safetensors and GGUF formats, packaging with MLflow and version compatibility. For ML engineers and data scientists.

    Model Serving & Deployment Corporate
    Live Course
    5 modules
    View

Get in touch