Translation. Region: Russian Federation –
Source: Novosibirsk State University –
An important disclaimer is at the bottom of this article.
From October 23 to 26, the IT holding T1 held a hackathon for young professionals and students at Novosibirsk State University as part of the "Impulse T1" conference. Over 300 participants from across Russia, from students to aspiring developers, took part in the competition.
The event continued the series of hackathons previously held in Nizhny Novgorod and Yekaterinburg and became one of the largest platforms for the exchange of experience between students and experts from leading IT companies. The total prize fund for the entire series was 3.8 million rubles, 800,000 of which were awarded in Novosibirsk.
Participants worked on two case studies. The first involved creating a system for generating personalized backgrounds using an ML module that would run locally on the user's device without transmitting data to external servers. The second case focused on automated Git activity analysis for development teams. Automated analysis enables the timely identification of problems in development processes and inefficient resource allocation within teams.
Teams from NSU won all three prize places in the first case.
The "Pirates of the Ob Sea" team (Maxim Yemelyanov and Ulyana Zaitseva, master's students from the Faculty of Mechanics and Mathematics, and Mark Skvortsov, Alexander Shovkoplyas, and Alina Vdovichenko, graduates of the Institute of Intelligent Robotics) took first place. The team received a prize of 180,000 rubles.
"I led the team solving the case 'Digital Dress Code: A Background That Makes You Stand Out.' The task was to develop a model that segments individuals from their backgrounds in a video stream. The ML module had to be robust to low light, light interference, and similar colors, and the personalization module had to generate customized backgrounds based on templates and employee data, while adhering to contrast and privacy rules," explains Maxim Yemelyanov.
The team began their work by analyzing existing solutions and identifying factors affecting segmentation quality. They then studied open-source models, selected suitable options, improved them, and integrated them into the system. At the same time, they developed interactive elements over the video stream to clearly demonstrate the algorithm's operation.
"We didn't just aim to complete the task, but to make the solution as stable and visually understandable as possible. It was important that the model looked 'live' and responded to the user in real time. The challenge was that everything had to work locally, without cloud services, so optimization took a lot of effort," notes Maxim.
Alexander Shovkoplyas, who was part of the winning team, also shares his impressions:
"The main challenge for us was working with technologies we'd never used before—for example, running neural networks via JavaScript directly in the browser. We managed to not only create a truly high-quality solution but also convey to the hackathon experts the potential uses for it. We implemented features in the app that are useful for companies: their image is built on small details—a suitable background for video calls in line with their corporate colors, a convenient way to display employees' names and job titles on the background. This is precisely what we accomplished and expertly presented at the hackathon."
Second place in the same case was taken by the VI_KA team (Ivan Kadilenko, Maxim Unzhakov, Egor Palkin), and third place was taken by the UML team (Stavr Mariskin, Fedor Chumakin, Daniil Kolesnik, Damir Shaikhov and Yuri Malyugin).
Thus, all the prize places in the first case were taken by NSU teams, confirming the high level of students' preparation and their competitiveness in the federal arena.
The Impulse T1 hackathon became not only a competition but also a platform for students to connect with IT industry experts. Participants note that such formats provide an opportunity to apply their knowledge in practice, receive feedback from professionals, and see how ideas are transformed into real solutions.
The material was prepared by: Yulia Dankova, NSU press service
Please note: This information is raw content obtained directly from the source. It represents an accurate account of the source's assertions and does not necessarily reflect the position of MIL-OSI or its clients.