About Success Builder

How do you find your place in life? How do you find something to do that both comes naturally to you and makes you happy? The answer is that you have to apply the knowledge you’ve gained from university and from life itself correctly. The Success Builder Project features HSE University graduates who have discovered themselves through an interesting business or an unexpected profession. The protagonists share their experiences and lessons learnt and talk about how they’ve made the most of the opportunities they were given.

Ayagoz Musabaeva, who holds a master’s degree from the HSE Faculty of Computer Science, has successfully applied her research skills to business. After working in the laboratory of the specialised Institute for Information Transmission Problems (IITP) at the HSE Faculty of Mathematics, she went to work in the industry and is now conducting research in machine learning at Akvelon, an international company. In this interview with Success Builder, Ms Musabaeva explained how a data scientist can find herself in eSports and what to do if you lack soft skills in IT.

Have you always liked math? Why did you make it your profession?

Yes, mathematics was easy for me at school; my brain seemed to be able to solve problems on its own. That all changed at the institute because I had to learn how to study. That was great, because advancing in mathematics is very interesting. My choice of a university was based on an assessment of my abilities and the universities where I could study with such a background. There were only a handful of universities offering an education in fields in which I felt confident. I settled on the math department of the Kazakh branch of Moscow State University because I knew that specialists from Moscow taught there and that I could transfer to Russia to complete my studies. I would have gone straight to Moscow to study but my parents were against me leaving home immediately after graduating from high school.

After that, you decided to continue your education. Why did you choose HSE University?

I happened to stumble upon Konstantin Vorontsov’s course in data science. It seemed like a very interesting and promising industry in which I could apply my knowledge. At that time, there were only two master’s programmes teaching data science, and the programme at HSE University seemed better developed and more in-depth. More senior fellow students of mine from the MSU Faculty of Mechanics and Mathematics were also available to help me master this field and offer useful tips. I also liked the format at HSE University that combines modernity and an openness to experimentation with academic fundamentals while also maintaining a very applied focus. What’s more, HSE University is very supportive of its students; communication between teachers and the student community is highly developed here, creating an environment in which you can develop.

What do you learn in data science and who studies it?

The master’s programme at HSE University provides the necessary foundation to work in data science as a researcher and as a specialist. You can also choose additional courses and delve into the areas you like. I would say that data science can be divided into three areas: data analysis that trains analysts or data engineers; classical machine learning that trains data scientists who work with tabular data; and deep learning that trains data scientists who process images, video and audio. There are also separate divisions for narrower professions. I went into computer vision because it has broad applications.

What kind of academic preparation does the master’s programme offer and how did you make use of it?

The master’s programme enabled me, first of all, to improve my knowledge of Python and the main libraries for working with data analysis. At the Faculty of Mathematics, I worked with medical data in HSE’s IITP lab. My math background enabled me to understand the complex methods of processing medical images and my master’s studies helped me to code it all by strengthening my programming skills in Python. I found my first job in the laboratory. It helped me develop the ability to put my knowledge into practice, since real-world tasks are very different from what you study according to a plan. I had to work hard to convert theory into actual work tasks. The coursework covers the most interesting aspects of inventing models and the activity in the lab itself enables you to work with ready-made datasets, when most of a data scientist’s work is based on preparing data and setting tasks.

How did you wind up at Constanta? How did this influence your further development?

I was interested in digital analysis and vision research, so I looked for a job as a computer vision engineer. I posted my résumé and interviewed at several places at once. The Constanta recruiter found me himself. I had already received an offer from another firm, so I decided to go to the interview just out of curiosity. They turned out to have a great staff. I really liked the work environment and objectives so I accepted their offer with pleasure. Constanta develops websites and mobile applications for bookmaker companies. This company also had a small department of OSAI computer vision that did research projects for sports. Its tasks included automating the analytics of sports broadcast views and providing data on various sports. In this department, I was involved in automating the acquisition of data from eSports broadcasts for bookmakers, as well as creating algorithms for automatically collecting basketball statistics.

Working in a commercial company is very different from an academic organisation. I learned a lot in the industry. I learned about new tools, mastered the necessary approaches, strategies and practices. And in such a great team, it was easy to learn and work. It was my first job in a business; my hard skills improved a lot in terms of goals and applied tasks. I mastered many tools that are not required in the academic field and gained a comprehensive understanding of how these tools benefit the business. I continued to do scientific work because Constanta encourages research and initiates new products based on it, but the business requirements for research differ from academic demands. For example, just in recent years at conferences for industry representatives, there has been a requirement for implementation and proof of state for their articles. That is, now, when you’ve written an article, you must set out the working code and a part of the data on which people can reproduce your work, since the issue of application is very acute for many articles.

Did you continue your academic activities at the university level?

Yes, I was accepted to the IITP graduate programme and wrote several articles for Russian and international conferences. I was invited to present one of the articles at a conference abroad but couldn’t get a visa due to politics. I ended up leaving academics due to burnout and did not finish graduate school.

Shifting my focus to practical tasks really helped me recharge my batteries

I have gradually developed a healthier work-life balance by maintaining a proper relationship to work.

Data science is definitely a trend. How does the demand for data scientists affect such specialists and the field itself?

The demand for specialists in this field is still high. They’re looking for savvy, quick-thinking people, especially with an academic background. However, the industry has a lot of monotonous tasks. It is difficult to find a company with a wide enough range of projects and opportunities for experimentation and growth. In large companies with research departments such as Avito, VK, and Yandex, there is almost always a place for research and they encourage employees to write both academic and popular science articles for blogs and publication.

Generally speaking, data science and narrow specialisations primarily require specialists to have excellent fundamental knowledge. This is partly why there is a shortage of soft skills in the field. I have never seen work process organised in a clearly logical way in a single company with which I have dealt, either directly or indirectly. Because machine learning (ML) is still a fairly new industry for both business and developers, new tools appear all the time, and the time needed to learn, implement, and adapt these tools is not distributed according to a particular plan. Managers and directors have to constantly adapt deadlines to this uncertainty. In such conditions, it is rarely possible to work effectively and take on clearly-defined tasks.

Another fairly common problem is when good developers work in leadership positions. They are usually more interested in their own knowledge than in team development and management. And so I constantly encountered this lack of organisation. Unfortunately, along with all the advantages, there is this flip side of working in business. I have always looked for the optimum, so with the advent of more and more tools for optimising the development of machine learning, I would also like to see good management.

Where are you working now?

I work for the international company Akvelon. The company is mainly engaged in outsourcing, but also does in-house projects. What interested me about this company is the infrastructure for developing ML projects. For any ML project, you need data, and that data must be stored and versioned. You also need to store models and be able to compare them, as well as keep track of metrics and how the project is developing.

All ML projects require significant resources, both hardware and human, so there is a need to automate and simplify processes whenever possible in order to speed up the work. A growing number of tools are now appearing that can replace these procedural mechanics and that work in different configurations and conditions. So I would like to try to develop some of the newest tools such as iterative.io to provide the company with feedback and accelerate automation. Thanks to user feedback, you learn which things are most important for the digital community, what’s lacking, which situations are not taken into account, etc.

Will you continue doing research or develop your career in business?

At the moment, I can only do research as part of my business objectives. At HSE University, I was involved for a time in course management and promoting the IITP. It seems to me that it is important, first of all, to promote and develop science in the academic sense, making it accessible and interesting. I think that specialists and scientists can and should do something towards this end. I would also be interested in contributing to the popularisation of science.

There are many different challenges in the industry now because users have very high expectations regarding the speed and quality of various applications. This creates great prospects and stimulates the emergence of new niches in which ML and digitalisation in general can find their place, including for applied research.

I would like to improve my hard skills in this area and develop managerial qualities. My classical education in mathematics definitely helps, but I feel a need to take more in-depth basic programming courses. I am an inquisitive person. I plan to get my soft skills up to speed and supplement my engineering knowledge with managerial competencies in order to optimise management processes in business. The digital industry needs capable management. First, I think I will try to understand certain organisational shortcomings, to learn more about how the sequential implementation of projects works and why in some cases, as I have repeatedly encountered, processes are not organised properly.