Review of Udacity Machine Learning Engineer Nanodegree with AWS SageMaker
Introduction
I’ve completed this course and wrote review about it on another platform some time ago. This article was my most viewed article there, so I decided to share it here as well. I’ve checked the most recent syllabus of this course, and found that it has not changed much in the meantime. The core technologies involved are still the same, while projects and use cases are different. That makes this review still relevant and I hope it will still be valuable to those who are considering taking it.
This is a very short description of this course. After finishing it, I felt the need to share my experiences and to help other people that might be considering taking the same course. I hope that it might help someone to decide whether to take this program, how to finish it with success, get the most from it, and avoid mistakes that I have made.
What I write here is my own personal opinion and experience and of course, someone else’s might be different.
What is Udacity?
For those that have not heard about it yet, Udacity is an on-line platform for learning technical skills like: Data Science, Programming, Artificial Intelligence, Cloud computing etc. Machine Learning Engineer Nanodegree at Udacity is an extensive program that helps you to get ready for the job market in that field. It is a multidisciplinary program that teaches you many valuable skills like: Software Engineering, Python Programming, Cloud Computing and Machine Learning.
Ready to accelerate your Machine Learning and Data Engineering career? For those in our community looking to master these skills, I created a comprehensive guide: “Python Data Engineering Resources” — packed with 100+ carefully selected tools, 30+ hands-on projects, and access to 75+ free datasets and 55+ APIs.
Get a free sample chapter at futureproofskillshub.com/python-data-engineering. Available as ebook and paperback on Amazon, and as an ebook on Apple Books, Barnes & Noble, and Kobo.
Master the complete tech stack at futureproofskillshub.com/books — from AI to Python, SQL, and Linux fundamentals. Plus, discover how to maintain peak performance and work-life balance while advancing your technical career in “Discover The Unstoppable You”.
What can you learn in this course?
This course helps you to learn and understand how to do real machine learning projects combined with Cloud deployment in Amazon cloud. That is especially valuable for someone who is fresh from the university and has never worked as a part of the team. Understanding how the real life software development flow goes is the key to cracking the job market and becoming employable. For less experienced developers who might not yet get a chance to practice some of the techniques used in this course, it is a good introduction and starting point to learn more about: software testing, version control, deployment, code reviews etc.
Apart from that, in this course you will learn how to think like a Data scientist, how to ask the right questions and learn from data. It will teach you how to execute Machine learning projects step by step and how to improve results of your work. It will also teach you how to do all this in the Amazon cloud by using AWS services and Amazon SageMaker Machine learning platform to train, test and deploy your Machine learning models. Have a look at my Git repository of required projects for this course. My detailed report about the final capstone project can be found here.
A complete program syllabus can be downloaded from Udacity’s website.
Who can benefit from this course?
This was one of the biggest revelations for me while I was going through the course. I realised that I was already familiar with most of the topics in this course. So I was obviously getting only a limited value from it. Then I asked myself, who could benefit the most from a course like this? If you find yourself in one of the categories listed below, then this program is the right thing for you:
Data scientists looking to improve their SW development skills or Cloud skills
SW developer looking to enter the Machine learning field and learn about AWS
Junior SW developer or Data scientist interested in cloud services, especially AWS and Amazon SageMaker platform
How long does it take?
The course website states an estimated time of 3 months at 10 hour/week. In my experience this is a very optimistic estimate. It might work like that only if you can commit a few hours of work every single day. Ideally you would want to work on this course 6–8 hours a day, to finish it as soon as possible and avoid paying higher fees than necessary.
In my case, it took me more than 6 months to finish, and two attempts. The mistake I made was to underestimate the amount of work needed for the final project. The main reason was the data cleaning necessary to perform the modelling. The provided datasets contained a large number of inconsistent data samples. For example: character values within numerical features, mixed decimal and integer values, large number of categorical features that look like regular numerical features etc.
It was possible to perform the correct data pre-processing only after reading and studying thoroughly additional metadata documents, describing dataset columns in detail. Since the data sets had a large number of columns, it took quite a while to go through all of them. This situation was actually very good to have, because it is exactly like in real life. We very rarely, or almost never, get really clean and tidy data to work with. Most often it comes down to the 80–20 rule, where 80% of time is spent on cleaning and preprocessing the data, and only 20% is spent on actual model training and development.
How much does the course cost?
This is not one of the cheapest courses on the market. For that reason it is wise to make sure that one will have enough time to commit to it, before starting it. There are two regular options with different prices:
“Pay as you go” option with monthly fee $249 or
“Pay upfront” for 4-month access with 15% discount and total fee of $846
There is also a third option, when Udacity is running a promotion and gives a discount of about 60–70%. In that case you might get a “Pay as you go” option with only $99 per month which was the option I used.
Was it worth it?
Short answer: yes, definitely. I learned a lot about some great use cases and problem domains from the Machine learning area. The course structure and content are very well aligned with real-life projects and necessary skills. I only wish I had taken it more seriously from the beginning, so that it did not take as long as it did to finish it.
I hope you enjoyed this article and found it useful. If that is the case, please subscribe to receive more content like this, every week thank you!