All courses

Advanced · module A13

MLOps, Deployment and AI Infrastructure

By the end: Deploy, monitor and scale ML and LLM systems, with automated pipelines and cost control

Start here
Covers
Taking models to production and keeping them working at scale
Contents
4 chapters, 17 topics
Time
About 4 hours
Level
Advanced
Your progress
0 / 17

This module builds on Supervised Machine Learning in Depth. We suggest looking at it first, but the choice is yours.

Chapter 1 Python Engineering for AI

  1. 1.1

    Advanced Python for AI Engineers

    Decorators, generators, context managers and type hints

    Code: refactor a script with these features

    15 minBeing written
  2. 1.2

    Object-Oriented Design for AI Applications

    Classes, dataclasses and composition in ML code

    Code: wrap a model in a clean class

    15 minBeing written
  3. 1.3

    Testing, Logging and Configuration

    pytest, logging and configuration files

    Code: test and log a pipeline

    15 min Read
  4. 1.4

    APIs, JSON and Asynchronous Calls

    Calling external services reliably and in parallel

    Code: an async API client

    12 min Read

Chapter 2 Deployment

  1. 2.1

    From Notebook to Production Code

    Project structure, configuration and reproducibility

    Lab: refactor a notebook into a package

    12 minBeing written
  2. 2.2

    Serving Models with FastAPI

    Prediction APIs, validation and testing

    Code: a prediction API

    15 min Activity
  3. 2.3

    Containers with Docker

    Images, containers and reproducible environments

    Lab: containerise the API

    15 minBeing written
  4. 2.4

    Cloud Deployment and GPUs

    Cloud services, GPU instances and serverless options

    Lab: deploy to the cloud

    15 minBeing written

Chapter 3 MLOps

  1. 3.1

    Experiment Tracking and Model Registries

    Tracking runs and versioning models

    Code: MLflow tracking

    12 min Activity
  2. 3.2

    Data and Pipeline Versioning

    DVC, feature stores and orchestration

    Lab: a versioned pipeline

    12 minBeing written
  3. 3.3

    CI/CD for Machine Learning

    Testing data, code and models automatically

    Lab: a GitHub Actions pipeline

    15 minBeing written
  4. 3.4

    Monitoring and Drift

    Data drift, concept drift, alerts and retraining

    Code: drift detection

    12 minBeing written

Chapter 4 Scale and Cost

  1. 4.1

    Distributed Training

    Data, tensor and pipeline parallelism

    Reading and demo

    15 minBeing written
  2. 4.2

    Inference Optimisation

    Batching, caching, quantisation and compilation

    Lab: speed up inference

    12 minBeing written
  3. 4.3

    LLMOps and Observability

    Tracing, prompt versioning and evaluation in production

    Lab: trace an LLM app

    12 minBeing written
  4. 4.4

    Cost Management

    Estimating and controlling compute and API costs

    Exercise: a cost model

    10 minBeing written
  5. 4.5

    Module Review and Assessment

    Practical tasks and a scored quiz

    Graded deployment

    20 minBeing written