In a previous post, I’ve detailed and defended why AI enthusiasts should consider getting the TensorFlow Developer Certificate. Now that I have passed the exam and got my certificate, we’re doing things a bit differently.
I’ve read several resources where the authors shared their experiences regarding the exam. Yet, I still had some questions left unanswered. In this post, I’m answering these questions.
The main issues we’ll be demystifying here are:
- Why should you get this certificate? (very briefly since I wrote another post about this)
- I’m not comfortable using PyCharm, can I use another IDE during the exam?
- Can I use Google Colab’s GPU resources to speed up the training of my models?
- What are the resources that you need to study to successfully pass the exam?
- My thoughts about this whole thing
Ready? Let’s handle this.
PyCharm vs Google Colab
Maybe the number one headache I had regarding this exam was the fact that we have been forced to use PyCharm. The trouble created by this condition was twofold. First, I was not using this IDE in my daily-life coding, therefore it was one more thing to learn in addition to the certification program. Second, I couldn’t rely on my laptop for training models when there is a time limit.
The good news is: You don’t have to.
PyCharm is mainly used to install the environment and the exam plugin. Once the exam starts, you can train your models wherever you want, using the IDE of your choice. You just need to save your model and send it back through the plugin when you’re done.
For me, I was doing all the heavy work on the free version of Google Colaboratory, and it was more than enough. Besides, to be honest, when I say “heavy”, I’m a little bit exaggerating. You don’t need hours to train a model; it’s just that Colab’s GPUs make life even easier.
Long story short, if you are not a PyCharm guy, and/or you don’t have a powerful machine to train on, don’t worry.
The exam is testing your abilities to build and train models using TensorFlow. As long as you upload the trained models through the plugin, feel free to use any environment you’re comfortable with.
TensorFlow Developer Professional Certificate on Coursera
Here, I’m not reinventing the wheel. Every single post or video about this certification says that you should complete the TensorFlow Developer Professional Certificate on Coursera before taking the exam, so DO IT. It is the best resource to prepare for the exam. Besides, Laurence Moroney is a great instructor.
Of course, I’m assuming here that you have a basic understanding of Deep Learning concepts, as well as a decent familiarity with Python and TensorFlow.
TensorFlow Developer Certificate: The exam
I’m not going into the details of the exam; but, as they mentioned in the TF Certificate Handbook, what you’ll be doing during the exam is coding, training, and improving models. This is not an MCQ exam. You will have 5h to complete the exam and it is more than enough.
You’ll be tested on the five skills below:
- TensorFlow developer skills
- Building and training neural network models using TensorFlow 2.x
- Image classification
- Natural language processing (NLP)
- Time series, sequences, and predictions
One last thing to be aware of is that the exam requires an identity verification phase. When you pay the $100 fee, you won’t start the exam immediately. This phase took about 2 hours for me, and I don’t know if it takes the same for everyone. Just keep in mind that you won’t start the exam straight away after paying.
My thoughts about the certificate
What I like most about this certification program is the fact that it is practical.
By completing the Coursera course, you truly learn how to use TensorFlow 2.x for building Deep Learning models, image classification, Natural Language Processing, and Time Series. The exercises you will encounter during the course will consolidate your Deep Learning understanding with practical training examples.
Finally, after passing the exam you get the TF certificate and badge to show on your resume, LinkedIn, and GitHub. You will also have your name among the TensorFlow Developer Certificate holders in the Tensorflow Certificate Network.
The exam itself closed in 2024; the Colab trick still describes how the practical test actually worked.