AI Olympiad winner develops country’s advanced technologies

Arseny Shakhmatov from Chelyabinsk improves Sber’s neural networks

Our country has been proud of nuggets throughout its history. This has not changed today: young talents are storming new professions with might and main. Even yesterday’s schoolchildren already occupy good positions in leading IT companies. Thus, 18-year-old Arseny Shakhmatov, a native of the Chelyabinsk Lyceum, is not only the winner of many programming competitions, but is now a full member of the Sber team.

Arseny joined the bank team at the end of this August. He was called to make diffusion neural networks. They are needed, for example, to create images based on your text request. For example, a user writes: “A car with airplane wings,” and the model draws a picture. And now the quality of the image, its compliance with the request, the elaboration of details – all this is now the work of the young genius and his team.

“In the sixth grade, I became interested in the topic and began studying what and how from videos on the Internet. Then I actively studied video game development using the Unity engine. They also used AI, and I also became curious about its work. When I reached a certain level, I began to participate in competitions,” Shakhmatov spoke about his path to starting his career.

And he participated very, very successfully. He won prizes at two NTO Olympiads in the field of Artificial Intelligence, in Kaggle competitions and last year’s AIJ Contest, the winners of which, by the way, will share more than 11 million rubles this year. The competition takes place as part of the international conference on artificial intelligence and machine learning AI Journey. Of course, after so many victories, the tech giants could not help but notice the young talent. Now Arseny is working on “educating” the Kandinsky neural network.

“Initially, pictures and descriptions of them from open sources were loaded into the first version of Kandinsky in order to train it to recognize pictures. We encountered a problem where the text request does not match the image and the neural network receives an incorrect example. That’s why in the first generated pictures people, for example, could have 6 or 10 fingers. Now the task is to retrain the neural network in more detail so that the current generation of diffusion models understands the image correctly. This is our task,” shared Arseny.

Denis Dimitrov, executive director for data research at Sberbank and head of Arseny, believes that in recent years the portrait of a young employee has undergone significant changes under the influence of technological development and changing requirements for the working environment.

“Today’s young employees were born and raised in the digital age, making them “digital natives.” They are more likely to use new tools and platforms, and this greatly impacts their productivity in an environment where technology plays a key role. For example, the level of AI competitions for schoolchildren and students in terms of the complexity and importance of the problems being solved (and, most importantly, the quality of the solutions proposed by the participants) is already not inferior to the level of “adult” competitions,” Dimitrov said.

Of course, young men and women with similar talents have a wide choice of future employers, and big tech can offer significant resources for personal and professional development. In addition, they often give access to their latest developments. Such companies always offer clear career paths and opportunities for internal advancement. Another important factor is that working in a well-known company can be attractive and prestigious, which greatly affects self-esteem and recognition.

“Young professionals are looking not only for a place to work, but also for an environment that will contribute to their professional growth and personal development. An environment that will correspond to the values ​​and ideas of the children about a quality life,” concluded Dimitrov.