How to earn the AWS machine learning specialty certification

The Certified Machine Learning - Specialty (MLS-C01) certification by Amazon Web Services (AWS), validates your ability to build, train, tune, and deploy machine learning models using the AWS Cloud.

This exam tests one’s understanding of machine learning theory. You do not have to be a very experienced AWS developer who knows all AWS products’ full depth and usage. However, it is vital to have an in-depth understanding of AWS and the cloud.

Exam structure

1. The exam is three hours long.
2. The exam consists of 65 multiple-choice and multi-selection questions.
3. You will be notified whether you passed or not once the time is up.
4. The exam is divided into four domains.

Let’s have a look at four domains in detail:

1- Data engineering

This domain covers 20 percent of the exam. It covers a few AWS services and streaming tools, such as Kinesis Firehose, Kinesis Data Streams, Kinesis Analytics, S3, DynamoDB, RDS, and other products in the AWS Analytics stack like Glue and Athena. Understanding these tools in detail may not be important; however, you are expected to know what and where they are used.

2- Exploratory data analysis (EDA)

This domain covers 24 percent of the exam and includes some basic tools, such as data cleaning and feature engineering. This module will ask you questions about one-hot encoding, normalization, handling missing values, data visualization, etc.

3- Modeling

This domain summarizes 36 percent of the exam and is one of the most important parts. You must have a key understanding of SageMaker algorithms and their uses. Moreover, you should know how to incorporate other frameworks, such as Spark/Tensorflow, into SageMaker. This module also tests machine learning theory outside of AWS. Therefore, knowledge about model tuningConcepts like overfitting and underfitting, and concepts of regularisation, dropout, learning rates, and gradient descent should be thoroughly revised along with their implementation. Moreover, you need to understand model evaluation techniques and model metrics (such as accuracy, sensitivity, specificity, recall, and F1). Lastly, you should be able to apply optimization techniques for deep learning models.

4- Machine learning implementation and operations

This domain sums up 20 percent of the exam. It covers a few AWS servicesSuch as Lex and Comprehend, and you must know how to deploy models from SageMaker.

A few resources

Take a look at the following resources while preparing for your exam:

1. Official exam guide
2. Amazon Sagemaker
3. AWS Machine Learning Blog