FI graduate optimizes Amazon warehouses directly from Luxembourg

Author: Klára Petrovičová for fi.muni.cz

Oliver Roch founded his first startup in high school. The company has been running on its own until today, but Oliver only consults remotely from Luxembourg, where he works as a data scientist at Amazon. After studying computer science at FI, he changed his major and country and started business school at Bocconi, during which he had the opportunity to attend Erasmus at ETH Zurich. He told us more about his journey back to programming, interesting internships and studies in an interview with FI.

How did you get into studying computer science and specifically at FI MU?

I've been interested in programming since the end of primary school, and in about 7th grade I had a dream to become a hacker. I wanted to "hack" the online games I was playing at the time, and when I googled how to do it, I learned that I needed to know how to program first. That motivated me to program even more. I knew that FI MU was one of the best schools in Czechoslovakia and that there was freedom of choice of subjects. I also already knew some people in Brno and knew that Brno was a nice student city.

What major did you choose at FI?

At the beginning I wanted to major in web technologies, but during my studies I became more interested in mathematics. Gradually, I took courses on artificial intelligence as well, and I profiled myself in this direction until the end of my studies. I didn't want to go completely in the research direction, but rather in the practical direction. That's why I decided to go to Bocconi University in Italy, where they offered a course in Management and Innovation of Technology. I also wanted to learn the business side of things, that is, how to translate this theoretical knowledge into practice and create a working product around it. I also had the opportunity to Erasmus at ETH Zurich, where they had a master's program for students with a technical bachelor's degree, where they taught us the basics of business, and how to put our knowledge into practice.

How would you compare FI MU and the school in Bocconi?

Both required a lot of theoretical study. Bocconi had a lot of math in common with FI, which students complain about a lot, but I'm grateful for that because now I use methods like linear models or multivariable calculus. What I would say has been a bit different is the university's connection with companies. Although FI also works a lot with companies, in Bocconi and ETH companies drove the curriculum to some extent. The companies see what are the important skills for the future, and they choose candidates based on those, and that's how the school sets what to teach. There are events with companies at FI too, but there was a lot more and more regularly. Another difference was the student associations, where students could join different clubs like Consulting, Data Science and so on, and had even closer contacts with companies.

The last thing I would say is that there is a lot of elimination at FI (laughs). It's like that sometimes in undergraduate abroad too, whereas in Master's it's harder to get in, but once you're there you know you'll pass the exams somehow. It's more like students try to get very good grades. At FI, students tried to pass (laughs).

The schools I would say are the same in content, but abroad it's more "student-driven" I would say. But in the final analysis, they both gave me a lot.

Didn't you miss programming when you were an economics major?

I did. But during high school, a classmate and I started a web development company called Progresio. I did that all the way through FI, even when I was abroad afterwards. I always did programming in my free time. What I use in my daily working life is mainly knowledge from FI, whether programming or mathematical. In business school I learned more how to transfer a product into practice. But in the end, I came back to IT again.

What was it like starting a company in high school?

The idea came about when a classmate and I were writing a SOČ in 3rd grade in high school. That's when we created an online tool to do surveys. We made it to the Slovak national finals and decided to take the project further. Moreover, we had a lot of people in the area who were interested in creating websites, so we decided to start a company and make them on commission. That was one of the things that led me to the business as well. For example, when we were talking to a client or recruiting new people to the company who we afterwards trained. At the same time, at FI I became interested in Data Science and Artificial Intelligence and needed some sort of link between those worlds.

Does Progresio still work?

Yes, it does. But my role is more as an advisor. I advise on the direction of the company or bigger decisions, but I don't get involved in the day-to-day projects.

Before you joined Amazon full-time, you were an intern there. A year later you also interned at Microsoft, how would you compare the difference between those experiences?

The two internships were very different. Amazon was more on the FI style because it was purely technical, while Microsoft was more business oriented. Both were about AI and Data science but from different directions.

At Amazon, I was right in the warehouse, where I had to develop a machine learning model to predict how efficiently pickers were picking shipments and suggest optimizations based on that. I was right there and was able to physically observe the employees and decide what was good data to predict based on. I could also see the impact of my job and that it's not just some numbers in the background, but that I'm handling physical packages that are being sent all over the world. I only had two months for the project, so I had to learn how to effectively develop the model, iterate and change it, while also breaking the problem down into steps well at the beginning to get it done.

At Microsoft, it was more of a tech-business internship that didn't involve programming, but I was to explore the Czechoslovak market with the potential of data and AI products in the Cloud. I identified opportunities where these technologies could be used, especially in large companies. I was tasked with finding the main problems in these companies that could be solved using data science and AI, and then contacting Microsoft partners who are developing on their Azure Cloud and advising them on what technologies they could use. So it was more about market research.

Both internships ended up teaching me something different. As for the first internship, at FI we always focused on models and their architectures. At Amazon, I learned that the model itself is about 10% of the time and 90% is spent with the data, for example, getting the data, cleaning it, etc.

In the end, you chose Amazon for your future career, right?

I guess I was more comfortable with the culture of Amazon. What I liked during my internship was that the company tries to do everything based on data. In most companies, you make a PowerPoint presentation and present the results, but that presentation is very dependent on what kind of speaker you are and how you can sell your work. Whereas at Amazon, they present pure data - they write a so-called "white paper" that has to fit into 6 pages, then people read it during the meeting and ask questions. So the focus goes directly to the data and primary information. There is also a lot of freedom in the projects, which there are a lot of, and you can decide what to work on and how to work on it. Microsoft also has a broad base for people starting out in their careers, for example strongly supporting technical education in the cloud.

But what I liked even more was Amazon's innovation and the way it solves problems. Especially the fact that all decisions are made based on data and you don't have to worry about failure. On the contrary, the company motivates you to do it, because then you are not afraid to do new innovative things with uncertain results. So I chose Amazon to start my career.

What are you doing now at Amazon?

I'm currently working as a data scientist for Supply Chain Automation. We're developing optimization models that optimize how much inventory we have in each warehouse and how much orders we ship from where, so that it's as fast and cheap as possible for the customer in the long term. This is for warehouses across Europe.

How did you get directly to Luxembourg?

Here in Luxembourg is Amazon's European headquarters. When I was in Prague at Amazon, I got positive feedback, but unfortunately there weren't many other technical projects for me there. So I got a recommendation from the Czech CEO to Luxembourg, where we found a position in Supply Chain.

Can you imagine yourself moving into a management position in the future?

Yes. I would still like to spend some time gaining experience in technical things and go deep. But eventually I'd like to move into a position like Technical Product Manager or Data Science Manager, where my previous business and technical knowledge would allow me to be someone who can tell what problem is worth solving and what model is ideal for doing so. Often you have to take into account who will be the user of the model, whether it will be used by a business manager who will only use the outputs, or another data scientist, or whether it will be a model that automates the final decisions directly, because different models need different treatment and some have unwanted side effects.

Is there anything you would do differently in your studies and career path?

Honestly, probably not. I think I've always been heading roughly where I wanted to be, although the exact destination has always varied a bit. I probably would have taken more math courses at FI. A lot of the courses weren't required, so I took more applied machine learning courses. But, for example, I had to catch up with statistics on my own when I started working in that field. In retrospect, I would have liked to be even more involved in the student clubs at Bocconi. Back then, I worked very hard to get good grades, which didn't do me much good afterwards (at least not in my master's, because in my bachelor's I also got into Bocconi and ETH thanks to them). At the business university in Bocconi, contacts were very important, even for the future. You know who to call, and you have classmates who are founding successful startups or working in big tech companies.