Why join AIDAO

AIDAO’s first season brought together over 2,500 participants. After the online selection round, 30 teams proceeded to the finals, including AI Capybara, the team of Daria Ledneva and Timur Ionov, Master’s students at ITMO’s Institute of Applied Computer Science. In the 32-hour-long offline round, the students trained an AI model on 200,000 photos provided by Yandex Taxi. As a result, the model learned to detect various damage types, such as broken windows, dents, or scratches outside and inside the car, to assess a vehicle’s condition before the start of a trip.

“We had quite a dramatic participation story that showed us that in IT it’s important to take risks and think outside the box. During the competition, there were two leaderboards: one was public and one was private – this one was used to select the winners. Sometimes, teams train their models for the public ranking to get a higher mark, but there are always risks that such solutions won’t work as well in the private ranking. One way to avoid this is to connect several models together in an ensemble, and thus ensure the solution’s stability. We were following this strategy but it turned out to be ineffective. Three hours before the finals started, we changed it all, ditching the ensemble and training one large model instead. This is what led us to the victory,” shares Daria Ledneva.

The team’s solution proved the most effective and accurate at detecting damage, topping the leaderboard. Daria and Timur received their 600,000-ruble prize at the Yandex headquarters. As they both admit, this experience proved to be valuable: Daria deepened her knowledge in computer vision and continued learning more about it, while Timur gained more points for his portfolio, which allowed him to enter a PhD program at ITMO.

Here are some tips from the winners for those who plan to compete in AIDAO 2026:

• Broaden your horizons. For instance, you can find new methods for your solutions in the Hugging Face community Daily Papers, which puts together a daily digest of fresh AI papers, as well as at top conferences like NeurIPS, ICML, ICLR, ACL, AAAI, and EMNLP.
• Join professional competitions and hackathons. You can find contests like AIDAO on Kaggle and AIcrowd. Major Russian companies like Yandex, VK, MTS, T-Bank, and others also host their own hackathons.
• Make sure to gain a good understanding of the concepts and methods of deep learning and ML. According to Daria, it’s impossible to be a jack of all trades in IT, but if you know how a particular model works and what tasks it solves, this model can be adapted for any case.

“At the competition, I relied on my soft skills – generating ideas and prioritizing tasks in experiments – as well as hard skills: correct execution of ideas, data analysis, AI model validation, and interpretation of acquired results. After AIDAO I realized that the most valuable experience for me was taking responsibility for engineering solutions, working under deadlines and uncertainty, and communicating with my teammate. These skills are equally important for work in research and industry,” says Timur Ionov.