From the Source: How Marvin Bertin Found a Passion for Data Science While Studying Robotics

Screen Shot 2016-07-28 at 1.16.35 PM

I’m from France originally, and then I moved to Montreal, Canada to do my undergrad in mechanical engineering. For a very long time I was interested in mechanical engineering, aeronautics, things like that. Then I got very interested in robots after an internship in Germany, so I started looking around for a master’s program in robotics.

Robotics is such a cross disciplinary field, I wanted to focus on an aspect of it that would be the most significant in our generation. For example, the reason I switched away from mechanical is I feel like the time to be a mechanical engineer was during the industrial revolution. What’s really stopping robotics from being very significant is not the mechanical or electrical sensors and motors, it’s the software. Like, self-driving cars. We have all the sensors to make a self-driving car. The only thing stopping us is building software that is smart enough to make that hardware useful.

I started teaching myself programming during my last year at university. Programming small drones, and in that process I stumbled on data science. My initial appeal toward data science was because I wanted to work in robotics, but that evolved into being generally interested in data science and machine learning.

A photo posted by Galvanize (@galvanizehq) on

The reason Galvanize and University of New Haven partnered to created this one-year master’s program in data science is because they sees this huge gap in skills and a huge demand for people who have skills in data science. I came into this program really wanting to learn those new skills that are extremely relevant for our time. Specifically, I came in very interested in deep learning, which you could say is the A.I. branch of data science. It’s software that writes its own software. We call it machine learning because as we give it more data, it fixes itself. It learns from its mistakes and learns to improve upon itself, which is a very different paradigm from the old programming we’re use to.

I realized that machine learning could be applicable in so much more than just robotics. Right now I’m in healthcare. It’s so funny, it’s a field I would never imagine myself in. I’ve never had a background in biology. But it turns out if you have good skills in machine learning, you can make meaningful impact in almost any industry.

I was born in France, and I’ve lived abroad most of my life. My dad was an expat, so I’ve lived in India for a couple of years, Egypt for a few years, France, and then I moved to Canada for my undergrad, then went to Australia to study aircraft structures for a year as an exchange program. I finished my undergrad in Canada, and then came to Galvanize to do my Master’s degree. It was my first time in the US, but I lived four or five years in Canada, so it was a smooth transition.

When I graduated, I was really excited to go straight into industry and get involved in the tech industry and get involved in building stuff. But at the same time, I didn’t feel like I had all the skills I wanted. What I liked about Galvanize was it’s really a school that’s between industry and academia. The fact that I’m surrounded by all those startups in the same building is amazing. Within a few weeks of being at Galvanize I started working with a deep learning startup called Skymind. It wasn’t part of the program, but because the teachers saw I was so interested in deep learning, they said I could work with them if I wanted to.

What’s really nice about Galvanize is, there’s a curriculum, but there’s no limit to the extent to which you want to learn. You have a project at the end of class, but it’s just a guideline. You work on the project you want, and it’s all about how passionate you are about the project and how far you want to take it. On presentation day everyone has come up with so many different projects in so many different domains. Finance, chemistry, it’s really about what you want out of the program, not what the program forces you to learn.

The school doesn’t make you, you make the school. The only reason the school is any good is how much you’re ready to put into your education. The classes and teachers are tools to help you move forward, but at the end of the day you’re in charge of your own learning, and that’s what really makes Galvanize special.

I worked with Skymind part time throughout the program. I used it to teach myself another language, because the program was in Python but Skymind used Scala and Java. I decided actually to do the data engineering course, which was supposed to be in Python, I decided to submit all of my assignments in Scala.

I did my capstone project with Skymind, building a new kind of recommender engine using graph theory and word embeddings. What was really cool is we were able to present our findings at a conference called Data by the Bay. I was even given the main stage to talk about the algorithm we worked on for the past few months. It was an amazing experience, considering I’d come to the US basically a nobody, and suddenly I’m at the forefront of this conference presenting some of the research we had done.

Thanks to that presentation, I was noticed by a number of startups that contacted me afterwards and was able to get a number of offers before I even finished my classes. Driver, where I’m working now, is one of the ones that reached out.

Driver is a genomics company that sequences patient’s DNA and tries to extract information out of it. So the goal is to develop new types of lung cancer treatments. 10 years ago to sequence the human genome costed like a billion dollars. Now it costs only a few thousand. So suddenly you have this explosion of new data about patients, so there’s a burst of new startups designing new drugs based not just on the old way of doing research, but very specifically based on your own DNA.

The way we’re treating patients now is as if we’re giving one shoe size to everyone. Everyone gets the same drug, it doesn’t matter who you are. But with precision medicine and the advent of genomics, we can design specific drugs for specific people in order to reduce side effects and boost effectiveness to better levels. I was really impressed by the mission of Driver, so I’ve been working with them for almost two months now.

I was really excited to come to San Francisco because I love outdoor sports. On the weekends I usually go kitesurfing in the Bay, if it’s windy. I also like rock climbing during the week, and I basically bike everywhere I go. So mountain biking, and then I bike everywhere I go in the city. It keeps me pretty busy.

When people say follow their passion, I feel like that’s a bit too optimistic. It’s really about following what you like to do. Do it daily. Do it unapologetically. Only then will you grow into the person you want to be. Because passion isn’t something you find, but something that grows with you.

-Marvin Bertin, GalvanizeU Graduate in San Francisco, California. You can find him on LinkedIn.

Want to read more stories of our awesome community #FromTheSource? Check out our archives.


Level Up