Before entering Skoltech, I graduated from Moscow Institute of Physics and Technology with a bachelor's degree in Applied Mathematics and Physics, Faculty of Aerophysics and Space Research. I focused on Space Instrumentation, and worked on research projects to improve satellite communication systems and apply artificial intelligence to avoid transmission signal fault.

I first got to know about Skoltech when I attended some international conferences. According to the students, Skoltech gave them opportunities to learn from leading scientists about forefront research, international experience, access to a supercomputer, and much more. All this motivated me to consider applying to a Master's program in computational and data-intensive science.

In the last year of undergraduate studies, my friends invited me to join a team to participate in the NTI Olympiad dedicated to the Internet of Things (IoT). The Olympiad gives many opportunities, including admission to Skoltech. The first stage consisted of a bunch of theoretical problems. The second stage included solving non-standard tasks related to IoT technologies. As a result, I was fortunate to be accepted at Skoltech for the Information Science and Technology program*.

*Editor's note: Information Science and Technology MSc program has transitioned into the Data Science, Internet of Things & Wireless Technology and Advanced Computational Science.

Initially, my main goal was to get a better understanding of neural network architectures and, of course, gain practical experience. So, I joined the Joint Laboratory of Applied Research Skoltech-Sber (LARSS), which makes it possible for students to both learn and gain hands-on practice through an official internship at Sber. The work is built in such a way that there is continuous communication between students, a scientific advisor, and an entire team from Sber.

At the lab, I worked on various neural network-based models for scoring platforms. For example, my master's thesis is dedicated to multidimensional polynomial lattices for predicting the probability of a client's default. There were also other cool projects related to the development of AutoML, networks for transaction chains, macro analytics, and many more. All this work wouldn't be possible if I hadn't joined the joint laboratory, that's for sure.

Overall, the number of tasks is growing; all of them are challenging with the possibility of implementing personal and latest-article ideas. So, I highly recommend LARSS to upcoming students.

My Master's program includes various courses in High-Performance Computing with access to the Zhores cluster to carry out personal computations. The knowledge that I have gotten in the classes helps to parallelize computational processes more efficiently and reduce time expenses. Besides, many hard skills obtained for conducting research on the cluster, with Open MPI, CUDA, and other fancy instruments for organizing research helps me on a daily basis.

*Editor's note: Zhores is a flagship petaflops-level cluster "Zhores" is one of the most powerful computers in Russia specially designed for data-intensive computational modeling and Artificial Intelligence. Skoltech Master and PhD students can get access to the cluster for their research.

Now I am working on risk modeling-related projects. Particularly, my thesis is devoted to multidimensional lattices. The main problem that slows down the integration of neural networks into production systems is a lack of interpretation. For the credit bureau data, one way to solve the problem is to enforce monotonic constraints on an objective function. The polynomial lattices perfectly cope with the task and give state-of-the-art results, but have certain drawbacks such as scalability. In my thesis, I propose possible ways to avoid the existing shortcomings. Another two projects are the development of models based on macro-factors for better risk management in crisis periods, and the implementation of metric learning on clients' transaction histories.

In the near future, I would like to continue working as a machine learning developer for solving industrial and critical global issues. The major is associated with applying AI to impact corporate sustainability activity on a large scale.

We are Skoltech. A new type of university in Russia, established in 2011 in collaboration with MIT with the vision of being a world-leading institute of science and technology. From the zero stage, Skoltech has rapidly advanced along the way towards a top-100 young university recognized globally by Nature Index.