Thread by @svpino, I am not particularly inclined to Math.I do not have a Ph.D.I [...]

I am not particularly inclined to Math.
I do not have a Ph.D.
I do not like to read research papers.

But I do make a pretty good living working on the Data Science/AI/Machine Learning field.

You can also do it.

Here is how I got here.

I started developing software very early and built my career as a Software Developer.

I feel that a strong foundation as a developer is a huge advantage in the field.

If you are exploring this path, do not skim on your Computer Science fundamentals!

In 2015 I started taking classes in the field. Here are the relevant courses that I took:

- Machine Learning
- Reinforcement Learning
- Reinforcement Learning for Trading
- Computer Vision

Each one of these was one semester long. They gave me a good foundation of theory.

On my own, I took the following classes:

- Machine Learning ( @AndrewYNg - Coursera)
- Deep Learning ( @AndrewYNg - Coursera)
- TensorFlow In Practice (Coursera)
- Reinforcement Learning by David Silverman (YouTube)
- Stanford Computer Vision (Fei-Fei Li - YouTube)

I read books, watched many videos, and read many articles. I enjoyed very much Deep Learning with Python by @fchollet.

All of this has happened over the last 5 years and it's mostly the foundation of my knowledge.

(Yes, I had to read papers at school. But that's it.)

Approximately in 2018, I started transitioning from pure Software Engineering to a more Machine Learning-focused role.

The experience you get from "doing" is irreplaceable. You can't match it with all the courses in the world.

So early on, start creating things.

After spending a long time focusing my work on more "Data Science" tasks, I found that my sweet spot is right at the intersection of Machine Learning and Software Engineering.

Right away I started learning everything I needed to be effective in this area.

Day to day, my job varies, but it always centers around "productizing models".

Two words, a lot of work.

More specifically, I help answer this question: "How can we get this model that shows some promise and make it available for people to use in real-life scenarios?"

The demand for this line of work is virtually limitless at this point and the supply is severely limited.

It's a great time to be alive!

Every year we make leaps on what we can do with the hardware, but we still need a lot of people to harness all of that power!

If you want to get into this field, here are some tips: