I have been tinkering with the idea of preparing the TF Developer Certificate and I have finally decided to create some time for it. The main aim for this blog is to summarise the main concepts and the whole experience, alongside with providing pointers for preparing the exam.

According to the TF Certificate Candidate Handbook the followin are the main areas that will be tested:

1. TensorFlow developer skills
2. Building and training neural network models using TensorFlow 2.0
3. Image classification
4. Natural Language Processing (NLP)
5. Time series, sequences and predictions

On top of those, I will also explore TF-agents, the Reinforcement Learning library in TF, starting with some of the examples in the tutorial page. The main goals for this exploration will be to end up with:

1. A better understanding of the RL framework and in particular of the DQN approach
2. Implement an Agent able to play Texas Hold’em at a competitive level
3. Implement an Agent that day-trades on liquid ETFs (at a profit :D)

And finally, we will draw inspiration from the excellent posts from Lilia Weng’s Github pages blog.