A career boost

2025 marked the 10th National Technology Olympiad. Over this time, NTO has grown into an all-national competition that yearly attracts thousands of school students. For them, the competition is an opportunity to test their knowledge, take on challenging tasks based on industry problems, and visit some of the major companies in Russia. 

ITMO has been a part of the NTO for four years now. This March, the university hosted the final rounds of two tracks: Intelligent Robotic Systems (with support from the company ROBBO) and Big Data & Machine Learning (with support from Yandex, VK, and IT contest platform All Cups). 

In April, ITMO served as a venue for the finals of two other tracks: Infochemistry and Software Engineering in Fintech. The latter was supported by Sberbank, which provided data for the final task and invited students to tour the company’s Technohub and attend lectures by the experts from its R&D Center in St. Petersburg. 

“One of NTO’s main advantages is networking. Contestants can meet like-minded peers and make valuable connections. This is also a chance for them to join an exciting project and even get a job offer from an industry leader. At the event, they can not only turn to the experts with contest-related questions – but also receive recommendations from them on their future academic or professional paths,” emphasizes Vitaly Kartashov, a jury member of the Software Engineering in Fintech track and an engineer and a data scientist at Sberbank’s R&D Center.

Petrochemistry algorithms

The final round of the Infochemistry track brought together 28 school students from St. Petersburg, Moscow, Novosibirsk, Yekaterinburg, Vladivostok, and other cities. Over the course of a week, participants competed for the final prize both individually and as part of a team. The contest was held in two stages: students had to first solve chemistry and computer science problems and then develop a software solution. 

For the final task, students had to build an algorithm for predicting the properties of synthetic oil based on datasets provided by Gazprom Neft. Creating oil formulas is a complex, multistage process that takes several months. Hundreds of formulas need to be processed to develop a product with desired characteristics. Contestants needed to find a way to automate the process using advanced algorithms. 

The team IChemTeam, which finished first among the contest’s teams, proposed three solutions for the problem. The best was ranked their ensemble approach based on the autoencoder SMILES and gradient boosting, followed then by graph neural networks in combination with gradient boosting and an hybrid approach with the autoencoder SMILES and the random forest method. 

“We needed to tackle a large-scale task in a short time. That’s why it was crucial for us to properly define team roles based on our abilities. It was also of help that we all studied advanced chemistry and computer science in school and had extra classes in these fields, so while some contestants were baffled, we felt at home. I hope this isn’t my last time at NTO; I’m planning to join the contest next year and try my hand at other tracks, as well,” says Evgenia Chukarikova, a member of the winning team. 

Fight against fraud

The final stage of the Software Engineering in Fintech welcomed 50 participants across 12 teams from Moscow, Novosibirsk, Chuvashia, Tatarstan, Primorsky Krai, and other regions.

The teams were tasked to develop fraud detection software for Sberbank employees. The task was drafted jointly with the company’s Security Department, which also assessed the students’ projects.

The best solution was proposed by the team FinBoom. At the core of their technology is a fraud detection scheme that involves the self-trained ML model Dominant, which automatically generates the “portraits” of bank customers based on graphs and records anomalies in their behaviors. The technology notifies bank employees of suspicious transactions, which allows them to rapidly suspend operations in case of fraud. In turn, bank customers can file complaints about suspicious transactions themselves. Their request will activate data verification and anomaly search, after which the operator will be informed about possible scams. 

“What makes our software so special is that it’s fully functional. We’ve tested our system many times and each time we got accurate results. We stuck to the problem’s main requirements, and perhaps that’s what helped us win. Though we certainly wouldn’t come up with a quality solution in time if it wasn’t for our background. I already participated in NTO three times and even won it once. I’m going to take advantage of the perks I received and will enroll in ITMO’s AI Engineering program,” shares Nikolay Odegov, a member of the winning team. 

Winners and runners-up of both tracks received seven extra points when applying to ITMO’s programs. Additionally, all participants were handed gifts from the track’s partner Sberbank; they were invited to intern at the company within the next five years. And team Ber⁴ received special prizes for the best pitch.