Machine Learning Engineering & MLOps: From Model to Production

Course Structure 

• Duration: 5 days Executive Bootcmp (Including full Consultative Session with AI Experts) 

• Onsite: at specified VIP location 

• Include full Buffet breakfast and Lunch

• Access to the Catalyst AI Hub Africa Alumni Network

• 30 days of post-programme support

• 10 Engineering Labs 

• Projects:  2 production-grade ML engineering projects — a deployed ML API and a complete ML pipeline with monitoring

• Theory vs Practical: 30% / 70%

Ideal For

This course is designed for:

• Data scientists

• Software engineers transitioning into AI who want to build production ML systems

• ML practitioners and researchers

• Data engineers  

• Technical leads and engineering managers 

• Full-stack developers

• DevOps and platform engineers

Recommended prior knowledge: Comfortable with Python and familiar with basic machine learning concepts — able to train a simple model using scikit-learn or similar. No prior DevOps or cloud experience required.

Course Description

Participants will learn to build production-ready machine learning systems from the ground up — covering experiment management, feature engineering pipelines, model packaging, containerised deployment with Docker, CI/CD automation, and real-time monitoring for drift and performance degradation. The course every data scientist needs to become a complete ML engineer.

Summary of Learning  

By the end of this course, participants will be able to:

• Explain the complete ML production lifecycle and identify the engineering gaps that cause most ML projects to fail before reaching live deployment

• Instrument a full machine learning training pipeline with MLflow — logging parameters, metrics, and artefacts, comparing experiments in the tracking UI, and managing model versions through a structured model registry

• Build reproducible, production-grade feature engineering pipelines using scikit-learn Pipelines and ColumnTransformer — with automated data validation using Great Expectations at every stage

• Package a trained machine learning model as a fully documented, production-ready REST API using FastAPI — including input validation, error handling, health check endpoints, and API versioning

• Write optimised, multi-stage Dockerfiles for ML applications, build and test containerised model serving systems locally using Docker Compose, and push production images to container registries

• Build a complete GitHub Actions CI/CD pipeline that automatically tests, trains, evaluates, containerises, and deploys a new model version on every code commit — with automated rollback on deployment failure

• Detect and diagnose model degradation in production — distinguishing between data drift, concept drift, and system-level failures — using Evidently AI statistical monitoring and feature distribution tracking

• Configure production alerting for model performance degradation and build automated retraining triggers that respond to detected drift without manual intervention

• Deploy a containerised ML model to a live cloud platform with auto-scaling configured — managing compute costs, storage, and endpoint reliability in a production cloud environment

• Build and present a complete, end-to-end production ML system — covering pipeline, serving API, CI/CD automation, and monitoring dashboard — as a portfolio-ready engineering project