My Journey With ML — Part II/IV: The TensorFlow Developer Certificate — usefulness, preparation, and time commitment.
In March of 2020, the TensorFlow team released the TensorFlow Developer Certificate. This foundational certificate allows ML practitioners to prove that they have the skill set required to succeed in an entry-level ML role. For the uninitiated, TensorFlow is an open-source Machine Learning framework from Google. Apart from being a powerful numeric-computing software, ML developers and researchers can build and deploy production-grade ML models quickly.
In the second of this four-part series, I’m going to share my learning path with you so that you, too, can start from knowing nothing about ML and end with the skills to land an entry-level job in this space. I’m also going to share the advantages of getting certified, along with whether the certification will be useful to you and whether it’s worth your time and money.
What’s A Certificate Worth, Anyway?
There is a tense debate in the tech-world about whether certifications are useful on a job seeker’s résumé. Some people argue that they are utterly useless and that substantial projects, fantastic internship experience, and a proven history of success in an industry-based role are what count. Others believe that they help demonstrate that the applicant has had some experience with the domain and can boost the candidate’s employability. I don’t think either side is incorrect.
I believe that certifications do have their place. In my experience, getting certified at something has always opened doors of opportunity for me. For example, my University entrusted me with building multiple official apps and allowed me to start the Mobile App Development Club in 2018; that I had become a Google Certified Associate Android Developer was a huge credibility-booster in that regard. Ever since I received my TensorFlow Developer Certificate, my LinkedIn profile has been getting a tremendous amount of traction, with people from all walks of life messaging me to ask for my suggestions and advice. This certificate has also made for an excellent addition to my résumé. It gives me something to talk about with recruiters right off the bat.
Do I think certifications alone are enough? Definitely not. But I would be lying if I said that getting certified hasn’t improved the quality of my professional life. Are you wondering if it’s right for you? You will be far better prepared to make that decision by the end of this article.
Are You A Visual Learner?
Another heated debate in the world of tech is whether to study from textbooks or online videos. Some argue that nothing can beat the comprehensive, thorough nature of a good, old fashioned textbook. Others say that online courses and videos provide a modern, graphical, and personalized way of consuming learning material.
My views on this are unambiguous- I’m all for the latter. ML concepts tend to be best explained graphically because they involve a great deal of mental visualization to process. For example, I find it much easier to understand how a Convolutional Filter moves over the pixels of an image when demonstrated through rich animation, rather than in a static textbook, which leaves everything to the imagination.
Furthermore, as I had mentioned in Part I, this field moves way too fast. While an online course may be updated regularly to feature the latest content and trends, a textbook may find itself becoming obsolete only a few months upon release. By extension, most books available in the market today may have already become obsolete, or at least not contain the latest and most updated information.
“I Don’t Have A Clue Where To Start!”
Neither did I when I first started. As I mentioned in Part I of this series, there are too many resources available, with too little guidance on choosing. However, after having gone through the process and doing my research, I write this section, so I’m confident that this path will be the best for most beginners.
Please note that I am in no way sponsored by or affiliated with anyone; everything I recommend here is from my research and experience. Also, I am in no way responsible for your performance in the exam, and following the steps below is at your own risk.
Step 0: Learning about AI
When I began my journey, I skipped this step, something I regretted later. Please check out the AI for Everyone course from Coursera and deeplearning.ai. Something is very compelling about this simple, non-technical course that makes the rest of the journey so much smoother. I took this course later on, but it should’ve been the first thing I did in hindsight.
Step 1: Python
Are you comfortable coding in Python? Before I started, I was unfamiliar with Python and a little afraid of it. Coming from the Java/C++ world of brackets, the thought of having to learn the Pythonic way of doing things scared me. However, I took up the AI Programming with Python Nanodegree from Udacity and was left feeling quite confident (though it took a while to get used to). However, I would suggest considering the free Introduction to Python Programming course from Udacity, as it pretty much contains the same content covered in the Nanodegree. If you have any other preferred resources in mind, feel free to consider those instead. If you’re familiar with Python, you can entirely skip this step.
Step 2: Deep Learning Fundamentals
A common mistake I see many people make is that they skip this step and directly move to the next; thus, they develop little to no understanding of the underlying fundamentals required to appreciate Deep Learning. I recommend the Deep Learning Specialization from Coursera and deeplearning.ai. The trainer, Andrew Ng, is a brilliant instructor and manages to distill years’ worth of experience into a single Specialization.
In this beginner-friendly course, you will learn all the theoretical concepts (coupled with a generous serving of practical lab sessions) that will lay the foundation for everything to come.
Step 3: Applied Deep Learning with TensorFlow
Take the TensorFlow in Practice Professional Certificate (formerly called the TensorFlow in Practice Specialization) from Coursera and deeplearning.ai. This is the officially recommended resource on the TensorFlow Developer Certificate page. Here, you will gain excellent, practical, real-world, hands-on experience with TensorFlow and be adequately prepared to clear the examination. The instructor, Laurence Moroney, along with Andrew Ng, quite brilliantly covers the material in a way that will leave you feeling confident by the end.
The page also lists an alternative program, the Intro to TensorFlow for Deep Learning course from Udacity, but if you intend to give the exam at some point, I would not recommend this.
Step 4: The TensorFlow Developer Certificate
At this point, you should be sufficiently prepared to take up the exam; more on this in the next sections.
Step 5: A Few More Advanced Topics
This may be looked at as a somewhat optional step, but I recommend it. Take the TensorFlow: Data and Deployment Specialization from Coursera and deeplearning.ai. This Specialization covers specific topics, some of which may not be too relevant to you (like deploying models on the web browser, on mobile devices, and IoT devices), but contains certain portions (like building data pipelines and Federated Learning) that I would not recommend you miss out on.
So, What Will This Take?
- Time– I would, in general, recommend that you dedicate up to a month for each Specialization, with about 2 hours of study time per day.
- Course fees– You can choose to either audit the courses, whereby you will have access to the lecture material for free, but not get a Coursera certificate at the end; or pay for and take up the Specializations. The costs are, in my opinion, reasonable for what you get.
- Exam fee– At the time of this writing, the exam costs $100. Is the price justified for what you get? I believe so, but you will have to decide for yourself (more on this in the next section).
- Dedication– A lot of it. Especially if programming is new to you. Things break and stop working, and you may find yourself spending late nights debugging your code. The installation of software is, in itself, a bit of a challenge.
Some tips from my experiences:
- Read the TensorFlow Candidate Handbook multiple times to familiarise yourself with the exam. Please note that everything about the exam itself, beyond what is mentioned in the handbook, is confidential.
- I cannot stress this enough. If you’re running Windows 10, pause Automatic Updates before taking the exam. The last thing you need is for your computer to start running a background update once you start model training!
- Familiarise yourself with PyCharm before taking the exam. Ensure that you are comfortable with it. Run some of your code on it to be sure that everything works as expected.
What’s In It For You?
For starters, you get a certificate and a badge.
You also get added to the Certification Directory.
You will have something to talk about with recruiters, add to your résumé, put in your email signatures, and improve your LinkedIn profile. Your skills will have been validated for an entry-level role in this domain, and you will feel confident in your progress.
Please, however, note that the certificate, by itself, may not be enough to help you stand out; projects, internships, and work experience are for that. Furthermore, the credential expires within three years of your receiving it, and you will have to pay and recertify to retain the title. If you’re a veteran in the industry, this certification may not be useful to you.
If you choose not to take this exam, you may be happy to know that more advanced certification exams will be released in the future, according to the official website. As for me, I cleared the exam, and I haven’t regretted taking it even once.
I took another certification exam that may be of interest- the Google Cloud Professional Machine Learning Engineer examination. More on that in the next part of the series!
If you liked what you read, please share this piece. Subscribe to receive notifications about future articles. If you have questions, suggestions, or feedback, let’s discuss in the comments!