N a n o d e g r e e p r o g r a m s y L l a b u s

LEARNING OUTCOMES

LESSON ONE

Enabling Security

• 

Create a Service Principal account for different types of roles

• 

Determine what the differences are in various forms of       

  authentication

• 

Use a specific type of authentication when selecting 

  deployment settings

LESSON TWO

Deploy a ML model

• 

Use a production environment for deployment

• 

Enable authentication in the deployment cluster

• 

Discover the differences between container-based deployment 

   and kubernetes.

---

Machine Learning Engineer with Microsoft Azure |  5

Need Help? Speak with an Advisor: 

www.udacity.com/advisor

Capstone Project

The program capstone gives you the opportunity to use the knowledge you have obtained from this Nanodegree 

program to solve an interesting problem. You will have to use Azure’s Automated ML and HyperDrive to solve a 

task. Finally, you will have to deploy the model as a webservice and test the model endpoint.

CapstoneProject

 

Training and Deploying a 

Machine Learning Model 

in Azure

You will be using both the hyperdrive and automl API from 

azureml to build this project. You can choose the model you want 

to train, and the data you want to use. However, the data you use 

needs to be external and not available in Azure’s ecosystem. After 

you have chosen a dataset, you will have to import the dataset 

into your workspace. Subsequently, you will train a model on 

that dataset using automated ML and then train a custom model 

whose hyperparameters you have tuned using HyperDrive. The 

type of model you use is not important. You can use ML models 

through Scikit-learn or Deep Learning models like ANNs and CNNs 

through Keras, TensorFlow, or PyTorch for this part of the project.

After you have trained both the models, compare their 

performance, deploy the best model as a webservice and test the 

model endpoint.

This project will demonstrate your ability to use an external 

dataset in your workspace, train a model using the different tools 

available in the AzureML framework as well as your ability to 

deploy the model as a web service.