Recently, my colleague João Pedro Boufleur and I obtained the Machine Learning Specialty certification from AWS. In this 4-part series of blogposts, we will share some tips about the test alongside some services and concepts that we learned studying for the certification.
This first post will cover the basics of AWS Certifications and where to find material to study, whereas, in the next three, we will talk more about the specific areas of the exam.
AWS (Amazon Web Services) is one of the biggest on-demand cloud computing platforms. It offers more than 175 products and services including computing, storage, networking, database, analytics, IoT, and Machine Learning. To validate the expertise of cloud professionals in different areas, AWS has created a program of certifications composed of 4 levels: Foundational, Associate, Professional, and Specialty.
The Foundational level has only one certificate: the Cloud Practitioner. This certification is the first level and is aimed for individuals that show an overall understanding of the AWS Cloud, regardless of specific technical roles. The next level is the Associate, and it’s aimed at professionals that have at least one year of experience working with AWS in specific roles (the Associate level includes one certificate for three different roles: Solutions Architect, SysOps Administrator, and Developer).
The Professional level is a kind of “upgrade” of the Associate level, having two different certifications: the Solutions Architect and the DevOps Engineer (which combines the knowledge of the SysOps Administrator and the Developer). Finally, the Specialty certifications combine knowledge of AWS with domain knowledge not necessarily linked to AWS. Currently, there are six distinct certifications on the Specialty Level: Advanced Networking, Security, Machine Learning, Alexa Skill Builder, Data Analytics, and Database.
All the certifications follow a similar pattern in terms of testing: there are 65 questions, and the result is a score from 100 to 1000, calculated using not only the number of correct questions but giving more weights to harder questions (Find more here). To be approved, you have score 700 on the Foundational exams, 720 on the Associate and 750 for the Professional or Specialty exams.
The Machine Learning Specialty certification is designed for people who work in data science roles and use AWS. The main objective is to test the candidate’s ability to design, implement, deploy, and maintain ML solutions for given business problems. It consists of four areas, each with different weights on the exam: Data Engineering (20%), Exploratory Data Analysis (24%), Modeling (36%), and Machine Learning Implementation and Operations (20%).
In the test, some questions focus on using specific AWS services (like Kinesis or Glue) whereas others focus on machine learning models and techniques outside AWS. However, most of the questions focus on both areas at the same time: for example, given a specific business problem, what is the best ML solution to solve it and how to implement it using AWS.
In terms of materials of study, there are a lot of recommendations to be made. In terms of official materials from AWS, the best we found (and used to study) was:
– A generalist course on Machine Learning,
– The Deep Dive video series on Youtube,
– Ryan Ahmed’s Machine Learning Certification Exam;
– Frank Kane and Stephane Maarek’s Certified Machine Learning Specialty;
– Abhishek Singh’s Practice Exams;
– Frank Kane’s Full Practice Exam;
In the next post of this series, João Pedro will talk more about the first two areas of the test: Data Engineering and Exploratory Data Analysis.
Featured Image by Christopher Gower on Unsplash.