This is a short story of how I've "abandoned" a music career to jump into Data Science and also encourage others to do the same (if you feel like that's for you).

Early days

Well, it's impossible to start this story by not saying that I was a music fanatic since the childhood. But changes have always been part of my life. Everything started with a passion for the electric guitar and, two years before getting into the university switched for the piano. The real surprise is about music itself: I've always thought that would be my profession until my last days.
But there came reality! And I started to realize that working with music wasn't really for me. Playing all over the city, travelling a lot, late night work and carrying instruments. Definitely not my business! (I love playing in a studio, record my stuff, but that doesn't pay my bills)

Day Trading

During my master, I started to get really tired about music. But I didn't know what to do. So I had the great idea of becoming a day trader. And… it didn't work! And I don't advice trying day trading if you don't have other income sources. No, I didn't loose a lot of money on that, but there is so much pressure when you need to take money out of the markets. It didn't work for me. But when you deal with trading, you deal with numbers, graphs and … statistics. I say this period was my introduction to descriptive statistics. I dealt with subjects such as moving averages, standard deviations, Bollinger Bands and many others that are the core of what I do today (yes, I'm in a trading project). And there was a lot of predictive analysis there (but the algorithms were running in my head).

PhD

Well, if day trading was my first step into using data to predict an outcome, my PhD brought me inferential statistics. It's important to mention that I hated the methodology of my masters. I have transcripted and analysed tons of musical notes and I couldn't see any science in it. So, for my PhD I searched another area within Music.
That was Music Psychology!
And as the name says, it's psychology. And psychology deals a lot with experimental studies. And we need a lot of inferential statistics to conduct these studies. There was the time for me to discover ANOVA, t-tests, p-values and many other concepts that I deal daily today. Besides that, my thesis was about the main atributes to becoming an expert musician in the field of improvisational piano. And that's very similar to discovering feature importances or SHAP values. And I came to discover it recently.

Transitioning

In my PhD period I started to realize that I could have found a field to really jump in. I didn't know it's name. Then I discovered Data Science as a growing field that was really related to what I had been doing at the time. So I decided to become a Data Scientist.
Was it easy? No it wasn't and it still isn't. It's a constant learning process that I'm still on (and think it will never end). Python, derivatives, matrix multiplication, Docker, Unix and so on…
It really never ends!
I can see many disadvantages of not being a STEM graduate or something like that. But fortunately I could take advantage of many of my previous skills and domain knowledge to the game. For example, being a non native mathematician makes me much more empathetic to non technical people and brings me a big concern on simplifying the language for including everyone in the discussions. Besides that, conducting a research (PhD) brings knowledge about the scientific method, which is very important to the Data Science field. In fact, for me, every Scientist is a Data Scientist.

What's next?

Well, recently I've discovered that I'm a practical Data Scientist (or maybe a full stack?!?! - I really don’t like this label)
What does it mean?
It means that I like to handle modelling tasks or solving business problems, but I also like deploying things. So I'm pushing my career more and more into Machine Learning Engineering. I like to see the models running, even if it's only a MVP. I like seeing them getting out of Jupyter Notebooks. I like monitoring the data that is being brought into to the model, checking if there are drifts and so on. So, I'm really dedicating my studies to tools such as Docker, Kubernetes, Airflow because I really see myself taking care of that (too).