The winners of the international machine learning competition Data Fusion Contest 2026 have been announced

VTB and IT holding T1 summed up the results of Data Fusion Contest 2026, an international annual machine learning competition.

This year, the competition with a prize fund of 3 million rubles lasted almost two months – from February 9 to March 30 – and brought together 2.5 thousand participants from 188 cities and 21 countries.

As part of the competition, participants – specialists in data analysis, artificial intelligence and machine learning – solved three problems: “Guardian”, “Cyber ​​Regiment” and “Heroes”. In total, 17,471 solutions were submitted. The “Guardian” problem aroused the greatest interest, collecting 9,507 solutions, followed by “Cyber ​​Regiment” (7,480 solutions). A total of 1,066 teams took part in the competition: 574 in the “Guardian” task, 410 in the “Cyber ​​Regiment” and 82 in the “Heroes”.

The Sentinel task was devoted to classifying unconfirmed transactions of bank customers based on historical data. Participants solved a binary classification problem assessed by the PR-AUC metric. The winner was an IT specialist from Moscow.

As part of the Cyber ​​Regiment, competitors had to build a multi-label classification model for 41 financial products based on anonymized client data. The quality of solutions was assessed using the Macro-Averaged ROC-AUC metric. The first place was taken by a student from the National Research University Higher School of Economics from Moscow.

This year the special Companion nomination attracted special attention, within which the best open solutions were evaluated. ZThe Heroes challenge involved solving an optimization routing problem with time windows (VRPTW), where specialists had to create the best travel plan, taking into account given constraints. Great interest in the task on the part of young participants emphasizes their high level of involvement and training, as well as the availability of modern data analysis tools for beginners.

“Data Fusion Contest For VTB — this is not just a competition, but a full-fledged platform for finding and developing strong specialists. We see the level of problems and solutions growing, and this directly reflects the maturity of the machine learning market. Participants successfully completed cases that were as close as possible to real banking processes. Such initiatives allow us not only to find new approaches to working with data, but also to quickly introduce best practices into business. Separately, I would like to note the high interest from young people — this is confirmation that a new generation of highly qualified specialists in the field of AI is being formed in Russia. VTB will continue to invest in educational and technological projects to maintain this potential and develop big data technology,” said Maxim Konovalikhin, Head of Data Analysis and Modeling Department at VTB.