‘A data scientist can never stand still’

For Okko platform users, the decision of what to watch is based on an ideal set of recommendations and an accurate search. The well-coordinated work of that search function is led by HSE graduate and Senior Data Scientist Shukhrat Khalilbekov. In this interview with Success Builder, he explains how he chose a master’s programme, studied and worked up to 16 hours a day, and what a senior data scientist in a major company actually does.

How did you make the move from St. Petersburg to Moscow?

I decided to move and begin studying for a master’s degree because I didn’t feel confident in the job market yet. I chose a master’s programme that had a balance between applied subjects and opportunities to try my hand at academic subjects. I was considering becoming a teacher and a researcher.

I had a chance to study abroad but was late submitting the necessary documents. In Russia, I had a choice between two majors—Applied Economics at the New Economic School (NES) or Financial Economics with the ICEF programme at HSE University. It seemed to me that NES was more focused on academia and more suitable for someone going into research. But ICEF gave me the balance I was looking for, in which I could communicate with industry professionals, study applied subjects, and at the same time prepare for a PhD. I opted for ICEF.

I think HSE is the best university in terms of strong disciplines, industry experts and great teaching staff. HSE University even helps you find a job: after defending my master’s degree, a teacher called me. He was a member of the examination board and invited me to work in his startup on very good terms, but I declined. By the time I graduated, I had two years’ experience in data science analytics and the job offer was for a starting position. It’s just that I had done well defending my master’s and the teacher thought I was a good candidate for his project.

What got you interested in data science?

It happened while I was getting my bachelor’s degree. During my second year, they began offering a minor in Data Science and they presented it really well: they talked about the analysis of the Dota 2 game and the relationship between the sale of an item and its monetisation. That got me very interested and I took the minor. I was hooked, although we programmed not in Python, but in R and did everything at a very serious level because there still wasn’t a lot of hype around data science. Because I studied as an exchange student for a year in Sweden, I only completed one of the course’s two years. So the knowledge I gained was fairly superficial and general. When I returned, I realised that I really wanted to develop professionally in data science, but that I didn’t have enough knowledge.

Before that, I wanted to go into consulting or try my hand in IB, but having already become interested in data science, I wanted to incorporate the technical aspect into my economics education. At that time, all the Big Three companies began setting up separate departments for advanced analytics that worked at the intersection of conventional consulting and data science. So I focused on preparing myself to work in this kind of technological consulting and paid attention to what the requirements were, as well as to who the large IT companies wanted to see.

How has the ICEF programme helped you develop in data science?

At that time, it was only the second year that ICEF was offering a new course on big data by Fabian Slonimczyk. The seminars were conducted by Stepan Zimin, senior analyst at McKinsey Advanced Analytics. This was a very useful and well-put-together course in addition to the programme’s fundamental disciplines. For me, it became one of the ways to structure everything I knew and learned about data science. I had picked up a lot of the information in the course on my own and some of it I learned at work. The course helped me to form a single picture and to mentor less experienced students in the course materials. I was already working in data analytics at the time. This course could prepare a beginner for a position as an intern or junior at an IT company. Of course, it is based on mathematics and econometrics—that are the core of the ICEF and computer science programmes at HSE University and very important when getting started in the profession.

Did you manage to combine work and study as a master’s student? Is that a good idea?

It is possible to combine them, but few students can manage it in practice. You need sufficient motivation and ambition for it because you’ll have to work hard and sleep little. For those 18 months of classes, you need to be prepared to devote 15-16 hours a day to work and study, and to keep yourself in shape at the same time. It seems to me that focusing only on studies, as apart from what’s happening in the industry, would not be as effective for a student’s development.

For example, I developed my technical skills such as coding at work. ICEF helped from the financial and economics standpoint, because now data scientists are very valuable on the market who can not only develop a model, but also find a solution to a business problem using analytics, and at the same time evaluate the economic effect. It was an effective combination of different types of activities: at work, I developed the technical side on my own, and at the same time I gained an understanding of what is happening in finance and economics and developed a business sense. This helps quite a bit and makes you a versatile individual who can deal with both financial and economic issues and develop a model at the same time.

What does a data scientist do in consulting? What exactly is technology consulting?

Oliver Wyman had its own internal startup—a banking platform where I began my path as a junior in a small team consisting of a product owner and another data scientist. (Subsequently, several other great guys joined our ranks.) We developed models for assessing the credit risk of legal entities using publicly available data and sold this service to banks on a subscription basis. As a result, the bank could simply go to our website, enter the taxpayer ID number of any organisation in Russia and receive an evaluation of whether or not it was a reliable borrower, as well as detailed analytics showing why the model reached this conclusion—and all within the framework of the IFRS 9. We performed detailed analytics and formulated conclusions that influenced decisions on whether or not to issue loans. As a result, several tens of billions of rubles were issued through our platform.

The consulting side was that we interacted with the client and then, using the bank’s numbers and portfolio, proved how much he could save, how well our model worked with his data, and why it was worth buying our solution.

At EY, I worked more on pricing projects—that are more like classic consulting projects—and did a little NLP (Natural Language Processing). Consulting starts where you and the client discuss your problem-solving methodology and his expectations, then put this in digital language (business value metrics) to green-light the project and, once it was completed, measure its effectiveness. The technological side concerns which tools to use and how effectively the problem can be solved. As practice shows, a powerful model is not always needed to help a business and this is worth considering. The data scientist must think through and form the infrastructure of where and how this technical solution will be supported.

Can a data scientist switch from a financial or consulting organisation to one that is strictly technological?

Yes, it is entirely possible and depends on the role. Looking at analysts, a data specialist with a financial background is in more demand in IT than in other areas.

In consulting and financial structures, a data scientist is able to accurately interpret the value of a particular solution and how it is predicted. Here, interpretation is very important and it is usually through practice that one develops the skill to understand the relationship between metrics. Generally speaking, IT companies always need such specialists.

Structural thinking plays a role here. This can be developed on the job, but it is better to do as a student, while going through the formative process. As time passes, it is difficult to relearn and change your thinking. But you can always pick up technical skills quickly, especially when there are many educational programmes available.

How and where can a person study structural thinking?

It’s really a mixture of university and work. At HSE University, lots of math analysis and case-based learning puts your head in the right place. It was thanks to a lot of both at ICEF that I learned to think properly. A big advantage of ICEF is that it teaches students to endure all the difficulties of an academic and applied workload. This really helps you develop while a student. At work, if you take an entry-level position with a top company with great specialists, their mentoring and your own observations help you develop many qualities.

How did you wind up working at Okko?

The job at Oliver Wyman focused more on finance and banking. In fact, I learned everything there that could be learned within the scope of my tasks and required competencies. Beyond that, I would have had to move up or horizontally, but all within the same sphere. At EY, I worked in metallurgy and NLP where the projects are very long and it got rather boring just spinning my wheels.

After that came City-Mobil where I was an all-around analyst who did everything and helped different product teams do projects. Unfortunately, though, they decided to close and I had to look for other options, although I wanted to stay at City-Mobil because they had interesting tasks and a cool team. At first, I started thinking about going abroad, but this happened just as there was a large exodus of candidates from Russia. Employers began lowering salaries, so it was no longer advantageous for me.

In Russia, I considered large companies such as Okko and Avito and received offers from both. It happened that part of my team from City-Mobil also received an offer from Okko, and this decided me in favor of that company.

The tasks at Okko are new for me because I had never worked with a streaming platform and online cinema like that. I work with the analysis and customisation of recommendations. It is interesting from the standpoint of both the user and the technical complexity. Globally, there are two main areas of focus in machine learning. Recommendations refers to offering personalised content to each user. The search function provides the most relevant responses to a query. I head the search division, and this is a new challenge.

Through which job gradations does a data scientist advance at this company?

As elsewhere: trainee, junior, middle, middle plus, and senior. The senior assumes maximum responsibility in making decisions, sets tasks for his team, and considers how to improve the accuracy of the recommendation/search function. Further, the task of the senior is to propose ideas, start implementing them himself or with a team, and delegating, controlling, and analysing the work.

What would you say is interesting about Okko?

The company’s business is interesting in itself. I myself use this platform, which makes me doubly inspired to do something: to see on the screen the changes I’ve made, to know that right now many people are evaluating my influence on how Okko works. The company has a very good team. These are ambitious, open-minded people, excellent specialists. In general, the company cares about its employees and offers many different perks.

What are the prospects for a senior data scientist?

A data scientist can never stand still. Although there is an unchanging foundation in both university studies and at work, there are many other algorithms, approaches, and technical capabilities that are constantly being updated and applied everywhere. You also have to constantly update your own knowledge and take courses, and this is interesting. I am now taking two courses simultaneously and will take more later. There’s a lot going on in data science. You need to keep your finger on the pulse if you want to build a successful career. At the same time, I would like to be an all-around data scientist that can take on different tasks and areas of interest.

Going forward, I would like to do projects that have the greatest positive influence on society, to apply my knowledge while improving people’s lives. Such tasks are most often addressed at large companies, and to work at large companies, you need to be strong and develop.