Alumni story: Arron, musician and recruiter turned Data Scientist

When I was a kid, I learned basic programming to make terrible games (laugh) but eventually my interest in music outweighed it. It could be just my assumption, but I feel that musicians tend to have a natural intuition towards programming or any other written language. There's actually a lot of musicians at my current company! 

At university, I studied guitar and afterwards worked as a musician on cruise ships floating around the world. My growing interest in Asia brought me to Japan where I found a job as a tech recruiter for data science positions. Chatting with data scientists and product managers from firms at the forefront of technology sparked my interest in tech again. 

I started learning Python online, but it was hard to filter through all the info out there and identify the core skill set I would need for a career transition. I decided to search for on-campus courses with teacher support, and that’s when I found Le Wagon. 

Le Wagon Data Science bootcamp was definitely a great challenge! I really appreciated all the support throughout the course, and a really good approach on how to help students go through the tasks. Instead of offering us a solution on a platter, all teachers were great and gave us the capability to solve challenges by ourselves - that’s actually what I needed. In the development world, your code often crashes and now I have no problem scrolling through errors and looking for the correct solution. 

One of my favorite modules so far was Deep Learning. It was fascinating to see how deep learning solves problems differently compared to the more traditional scikit-learn methods, and how much flexibility you can get out of models. Our Deep Learning teacher Yann literally infected anyone with his enthusiasm and passion! 

    

I pitched my idea of solving an issue for musicians learning a new track and had two batchmates joining me. The goal was to run a song through a Deep Learning model and had it split into separate musical instruments. With the help of Aishu and Njeri, we came up with a solid-looking product that had a wow effect during our Demo Day presentation. 

It was really important to see other people's approaches and get a better understanding of how to have a business impact as a data scientist. Many people optimize algorithms and solve obscure problems on an academic level, whereas the bootcamp focuses on applying Data Science knowledge to real world problems with an actual impact.

Lots of people have the impression that data science requires a math background - that’s not true! My background is light years away from STEM. Of course, having some competency or at least the willingness to learn statistics and inference is needed to pick things up. But you are not expected to be a math wizard to become a data scientist unless you want to go into academia or work at the R&D department of a firm tweaking hyper parameters and building new architectures for models. The majority of data scientists are using already predetermined architecture. 

After graduation from Le Wagon Data Science bootcamp, I spent one week before Christmas holidays sending out as many applications as possible. When January came, I went through my emails and a few people had reached out to me already. It took me just 2,5 weeks after graduation to get a job offer from Hacarus, a Kyoto-based AI startup that just opened up a new office in Tokyo. 

Hacarus has a very interesting approach: it focuses on an explainable AI using sparse modeling and machine learning methods that can be visualized so that the end users can actually see what it's happening. In addition, the company has a really nice atmosphere, which was very important for me coming from a very high-stress environment. People here deliver great results because they like the products they’re working on, rather than have a supervisor breathing down their neck. 

I'm working on a computer vision task building and reconstructing different modules in PyTorch. It comes with an additional issue of how to effectively annotate and pre-process images to help model learn the features of the problem. 

Hacarus is a really enticing environment and a great place to learn new skills. We've got 15 or 16 people on the data science team, which means a lot of people to reach out to and learn from. 

Data Science and Data Analytics are the fastest-growing sectors for talent demand, and Japan has one of the most candidate short markets in the world. If you speak Japanese and English and you have a basic data science skill set, you will be literally hunted down by recruiters! I don't speak much Japanese yet I managed to land a Data Science job 2.5 weeks only after starting my job search. Keep digging, actively reach out to recruiters and hiring companies - and you will definitely find a great place to work.