Machine Learning, Education

How to Succeed at the TensorFlow Developer Certification exam (2021)

Resources to pass the TensorFlow Developer certification

Phani Rohith

Published in

Towards AI

6 min readFeb 27, 2021

TensorFlow Developer Certification tests user understanding of TensorFlow developer skills, building and training neural networks, image classification, natural language processing, time series, sequences and predictions using TensorFlow 2.x. Let’s look into the key things about this exam.

Sections covered:

1.Exam cost

2. Stipend

3. Length of validity of the certification

4. Who can take the exam?

5. Resources allowed during the exam.

6. Exam time limit

7. Resources I used to clear the exam

8. Results

Now let’s dive into these sections:

Exam cost: $100 USD (one attempt)

From the date of purchase, you’ll have six months to take the exam. After six months, your purchase will expire, and you have to purchase again.
If you don’t pass on your first attempt, you must wait 14 days before retaking the exam.
If you don’t pass on your second attempt, you must wait two months before retaking the exam.
If you still have not passed the exam after three attempts, you must wait one year before you retake the exam.

2. Stipend:

To ensure this opportunity is available to everyone, TensorFlow provides a limited number of stipends worth 50% of the exam cost.

Amount: $50, which is 50% of the exam cost.
Attempts: One certificate purchase includes one attempt.
If you are interested, fill out this application form, TensorFlow Education Stipend. Your selection will be notified with 4–6 weeks of submitting your application form.
Stipend expiration: The stipend must be used with 90 days of receipt. Once you purchase the exam with the stipend received, you’ll have six months from the purchase date to take the exam before it expires.
Note: You can only apply once a year if you are interested in earning a stipend. Please do not include more than one application a year. If you submit more than one order, in future years, they can no longer consider you for a stipend every year.

3. Length of validity of the certification:

Your certification is valid up to 36 months from the date you receive your digital badge. To renew your certificate, you need to write the exam again.

4. Who can take the exam?

This certificate is for students, developers, data scientists or Machine Learning Engineers who want to demonstrate practical machine learning skills on building and training models using TensorFlow.
As an individual, you must be tested, and this credential is not for businesses or legal organizations.

5. Resources allowed during the exam:

You can use any learning resources you would typically use during your ML development work.

6. Exam time limit:

Five hours: You’ll have five hours to finish the exam, and in case you click the submit button before you complete the exam, it will be auto-submitted and will be graded for the questions you submitted and tested models.

7. Resources I used to pass the exam:

a. TensorFlow Developer Certificate course by DeepLearning AI:

Assistance: 10/10 (Mandatory)

This specialization is the most useful resource as it covers all the essential concepts required for this exam. This course consists of videos that explain the concepts, references to excellent articles and papers, quizzes and the most crucial section, Labs, where they ask us to solve the code, which is the most interactive component of this course.

If I have to select the one thing which is the most useful resource for this certification, I would say that it’s this course and it will be sufficient enough to clear the exam.

Time Duration (Minimum of 4 weeks):

This specialization has four courses. If you already know MLPs, CNNs, RNNs, and LSTMs, you might not require four weeks to finish this specialization. If you are unaware of these concepts, you might need a minimum of 4 weeks to complete the four courses in this specialization.

Note: The duration to finish this specialization is varied based on the amount of time spent per day/ week.

b. DeepLearning Specialization:

Assistance: 7/10 (Optional)

It’s always important to learn the basics to lay a strong foundation. This course helps you to build a strong foundation in the field of deep learning. This course covers Deep learning evolution, improving the neural network’s performance using hyperparameter tuning, regularization and optimization. This course also presents the growth and need of convolution neural networks (CNNs), recurrent neural networks (RNNs).

c. PyCharm:

Assistance: 10/10 (Mandatory)

Unfortunately, you cannot use google colab to take this exam. The exam has to be given in PyCharm IDE. PyCharm IDE is the most popular IDE for python. For this exam, you have to install PyCharm IDE on your local device and make sure you have enough RAM to run the DL algorithms. To know if your system is capable or not, go to this link and install the Community edition of PyCharm. Try out neural network architectures like MLP, CNNs, RNNs, and if it takes forever (greater than 40–50 mins) to train, then you might have to find any other device with a greater RAM size, and if you have a GPU (VRAM), then you don’t have to worry at all.

Once you pay the examination fee, you’ll receive a detailed Handout with instructions on using PyCharm to take the exam. I highly recommend you go through all the sections in the handout.

Things to focus on:

PyCharm Edition: Make sure you install the same Pycharm version and python version presented in the handout; else, you might face issues downloading the plugin and starting the exam.
Troubleshooting: You might get stuck during the exam or even before the start, so I highly recommend reading this section before giving the exam. Due to confidential compliance, I cannot reveal where I was stuck and lost 30 -40 mins during the exam to figure it out. The solution for it was available in the Troubleshooting section of the handout. So, there might be a possibility that can happen with you as well. Then open the handout and drive to the Troubleshooting section, where you might find an option to resolve it.
FAQs: This section covers all the possible questions you might have on starting the exam and PyCharm general queries.

8. Results:

If you have cleared the exam, you’ll receive the certificate to your registered email address within 3–5 business days. Once you submit the exam, the scores will be updated in the candidate portal. If you have passed the exam, you’ll receive an email telling you to add to the TensorFlow Certificate Network. Here you can find a list of TensorFlow developers who have passed the certification. There’s an option to opt-out if you don’t want to add yourself to this network. If you don’t clear on your first attempt, you must wait 14 days to retake the exam.