Learn some maths

Some will say you don’t need maths to create something useful but you will not understand what you are doing if you don’t grasp the meaning of things like “stochastic gradient descent” or “entropy”. In any case, if you are not familiar with basic algebra like vectors and tensors you will have to.

Learn some languages

Python, R, C++ and Scala are the most used language in AI and being able to switch between each is an advantage. If you want to master Torch you will need Lua, if you want to dig into TensorFlow you will need to program in Python.

Learn some frameworks

There tons of AI frameworks in all directions; robotics, NLP, image analysis, neural networks. Take time to experiment and embrace the diversity, it’s a dense jungle of variety. Probably you also need to appreciate IoT, GPU’s and cloud computing as part of this.

Start small, go big

Play with basic things and well-known datasets, invent things and go off-road. Mastering things starts with very basic questions; what does it mean to recognize someone’s voice, what is an emotion, how do you detect an intention from someone’s questions…? Don’t get lost in technical details and keep your compass close.


Understand how concrete business needs make contact with concrete tech techniques. Understand the limits and opportunities of frameworks and business ideas. Integrate AI as another tool in your box (next to enterprise architecture, security and whatnot).

Go beyond

Fill a gap, invent something, pick up a challenge within your own capacity. Explore. Write about it.