Connecting the dots while learning machine learning

With the new year upon us, I’m taking the opportunity to share the highlights of the journey I have embarked on when I decided to switch professional paths a little over a year ago.

It started with one decision, certainly the hardest I’ve had to make in my life. The impact of this decision was crystal clear: I was about to renounce to all benefits of a cushy corporate position, leave a company that I had loved and with great people, and what I thought would be “starting again from scratch” (although in retrospect, I discovered that it is probably never really the case). So what led me to take that leap? Over the previous year, my curiosity and excitement for the field of artificial intelligence and machine learning had grown immensely, and I was now starting to see glimpses of a new purpose for myself. Following this path seemed to me like a promising way to bring more to the world, solve important problems and leave a mark.

So it is with all my strength and determination that I embarked on this exciting journey. At first, the amount of things I needed to learn felt slightly overwhelming. I knew for example that I would need to get serious at improving my programming skills, which had become rusty over the years. I was also determined to ‘get fit’ on the aspects of statistical modelling and was also very keen to start building a sound skillset for data wrangling, visualization, machine learning and deep learning. I had a clear map of the skills I would need to gain and three months before the new year to see how far I could go. Thus I created my own bootcamp: I worked diligently 50 hours a week for three months and I consumed as many MOOCs (i.e. online courses) as I could, each time incrementally targeting new skills in my list. I used a combination of Coursera, Udemy, DataCamp and Udacity. I was so amazed by the quality of some of these courses and felt priviledge to have access to all this knowledge and be able to consume it at my own pace. I also used several books as references to help me gain more depth when needed. Here is a short list of books that I found particularly useful :

1. Introduction to Statistical Learning, G. James & al., Springer
2. Pattern Recognition and Machine learning, C. M. Bishop, Springer
3. Deep Learning, I. Goodfellow & al., MIT Press
4. Learning Python, Mark Lutz, O’Reilly
5. Hands-On Machine Learning with Scikit-Learn & TensorFlow, A. Géron, O’Reilly

This first phase of learning went really well for me, and I got to a point where I became comfortable coding in Python, R and SQL and using the data science and machine learning toolkits and started undertaking my own personal projects. However, I eventually started to feel that there was more and more overlap between the online courses that I had already taken and the next ones I had on my list. I was also concerned that these courses would not provide me the same depth of knowledge and opportunities as a university degree would. As a plan B to my self-teaching strategy, I had applied for a Master’s at HEC in Business Analytics and Data Science just in case I wanted to supplement my learning. I got accepted and was even offered the L’Oreal Admission Scholarship, which was a nice bonus.

I eventually decided to try one semester to test my new skillset and see how much more I could learn. I was determined to do it diligently, so I took one extra course to challenge myself further. The course load ended up being quite challenging but acceptable and I was able to make the Dean’s Honour Roll, which gave me a lot of confidence in my ability to learn and master new skills. After completing this first semester, the course I found most challenging and the one I ended up learning the most from was the Applied Machine Learning course which I was able to take at Mcgill University instead of HEC. This led me to question whether the Master’s at HEC was going to provide me with the depth of knowledge I was looking for. Hence, I decided to decline the second half of my L’Oreal scholarship and change universities to pursue a Professional Master’s in Machine Learning at MILA, Quebec’s Artificial Intelligence Institute.

After having completed my first semester at MILA, I feel extremely blessed and at my place. I was actually surprised and admittedly very happy to discover a large number of individuals like me, who have several years of experience in the industry in another field (e.g. geology, engineering, finance, mathematics to name a few). MILA is one of the top schools in AI and brings some of the most talented individuals from all around the world to do research. As part of this professional Master’s, I will not undertake a research project, but I will be directly contributing to translating MILA researchers’ state of the art AI methods and models into the industrial world, via a six-month internship that I will undertake in the Summer of 2020. I’m thrilled and definitely looking forward to this internship opportunity!

For many, the idea of leaving a highly paid job and return to an entry-level status may sound financially unreasonable, if not completely ludicrous. For me, it is an idea that I take with a smile. The whole experience has been eye-opening as far as showing me things that we tend to take for granted while also revealing what leads to my personal fulfillment and happiness. Also, it made me realize that ego can sometimes stand in the way of an opportunity to renew yourself, seek a new meaning from your career or be challenged in new ways. Recently, I discovered that I’m actually only one example of a now growing trend of established professionals taking on “minternships” (i.e. Mid-Career Internships).

I am excited about 2020, and am sure that I will have plenty to share again at the same time next year!

I wish you all a happy new year! Cheers!