Key solutions presented at AIMission

• Russia’s first system for automated multiagent AI development that can replace an entire team of developers and architects. Denis Nasonov, the head of the Laboratory "Industrial AI,” pointed out that this system doesn’t simply “generate code” but implements the entire development cycle, assesses its own performance, addresses errors, and creates industrial infrastructure. Thus, the system increases the quality and accessibility of AI solutions while also decreasing development time and cost. The solution can find applications in various sectors of the economy: for example, in designing and maintaining internal platforms in telecommunications, creating information systems and performing data analysis in public administration, developing digital twins and monitoring systems in industry, or automating LMS in education.

• Prostor – a multiagent AI platform for urban planners. According to Sergey Mityagin, the head of ITMO’s Institute of Design and Urban Studies, in 2025 the system was improved: now, Prostor can help evaluate a development project at the scale of a specific locale, a city, and even an entire region. Depending on the availability and composition of the initial data, the system can solve various tasks: assess social, engineering, and transport infrastructure, the environmental situation, the potential for using the territory for different functions (housing, recreation, industry, agriculture), and model the effects of implementing various development scenarios. Moreover, Prostor can predict the expected load on the traffic network, as well as a territory’s investment potential. Thanks to this, the time for analysis and planning of territory development is decreased from several months to weeks or even days. 

• An AI assistant for natural sciences scholars that creates a network of AI agents for complex high-level tasks, thus facilitating research. This AI collaborator will help scientists collect information, analyze literature, verify results, and prepare a technical task description for next steps. Next, researchers and AI can evaluate the results together, suggest new ideas, or clarify the goals. Anna Kalyuzhnaya, a senior researcher at the Institute for AI, added that thanks to this assistant, the research cycle is accelerated by several times. The platform is a universal tool and has been tested on relevant tasks in chemistry, materials science, and clinical medicine. The AI collaborator is a continuation of another project, ChemCoScientist, a digital assistant for chemists.

• A system for assessing human capital in AI that helps calculate how many AI specialists are found in a company, city, or region and evaluate their experience. The new system will automate the collection and analysis of open information about specialists: their scientific publications, conference presentations, contributions to open-source code and AI personnel training, and media publications about products. For example, analyzing scientific articles provides information on the focus and results of the intellectual activity of a specific developer, as well as the scientific teams, key organizations, and co-authors they work with. Gleb Glukhov, a researcher at the Institute for AI, and Artem Beresnev, a senior lecturer at the Faculty of Applied Informatics, noted that the system is capable of evaluating the innovation potential of an organization or an individual; thus, it can identify the ability to create new ideas and technologies in AI – from research to practical implementations.

• A platform for development of autonomous AI computer vision systems for remotely piloted aircraft and drones. Sergey Ivanov, the head of the Laboratory “Intelligent Decision-Making Technologies,” explained that to be fully autonomous, drones need to not only analyze the objects around them and environments they are in, but also make their own decisions based on this data. The project centers on a software platform that’ll be used to solve computer vision tasks, specifically those relevant for drones; this platform can become a tool for a wide range of applied tasks: from pipeline monitoring and environmental control to agrochemical field treatment. Unlike similar solutions, the new platform doesn’t require any programming skills and uses verified domestic solutions (GNS platform by GosNIIAS, SMILE platform by ITMO, and image “annotator” by KT-Bespilotnye Systemy), as well as cutting-edge AI models for computer vision. Particular focus has been laid on ensuring the system’s compatibility with the domestic computational base – all software has been optimized to fit Russian-made processors and onboard computers. The National Technological Initiative Foundation supported this project with a 162-million-ruble grant.

“This year, the Research Center ‘Strong AI in Industry’ has won the grant competition within the federal project Artificial Intelligence for the second time, also becoming the only institution in St. Petersburg to join the project’s third wave. We continue developing strong AI and we believe that it can only be created with a phase change – many AI agents interact with each other, causing non-linear effects, and at some point AI becomes strong. The main issue is not in assembling several dozens of thousands of AI agents, but to make an AI capable of creating multiagent systems for different purposes and tasks. In order to do that, we need to replicate the functions of a strong AI, automate development, effectively build AI agents and ensure their security – these are the tasks being solved by the center starting in 2025,” shared Alexander Boukhanovsky, the head of the Institute of AI and the School of Translational Information Technologies.

Role-based competency model

This year, following an order from the Analytics Center of the Government of Russia, ITMO experts have developed a new Russian system for training specialists – the role-based competency model in AI. It brings together professional knowledge and skills that will train students not to fit one specific profession, but to demonstrate a number of competencies corresponding to different roles. With the role-based competency model, the university can communicate with business using a common language to develop high-demand educational programs and train highly qualified specialists capable of joining industrial work processes. Among such programs is AI Engineering. The new approach is different from the traditional one in the fact that these new Bachelor’s programs in AI are designed based on current industrial demands and not on the capacities of a university, explained Alexandra Klimova, the head of ITMO’s educational and methodological center “Artificial Intelligence.” ITMO has already shared its experience of building educational programs based on the role-based competency model with over 300 educators from 22 universities in the country.

Thanks to a systemic and complex collaboration with industrial partners at each stage of design and implementation of educational programs, ITMO has been able to develop a model of mass training of highly qualified IT specialists that has recently been honored with the RuCode award. According to Alexander Mayatin, the head of the Center for Top-Level Educational Programs in IT, this approach helped attract resources available only to the industry, including experts, cases, data, computational power, and infrastructure. As a result, it was possible to increase the efficiency of the Bachelor’s program Software Engineering. The program is now supported by more than 20 partners, including Yandex, Avito, Alfa-Bank, T-Bank, and others, while 107 out of its 130 lecturers of core subjects are practicing IT specialists. 95% of the program’s graduates secured positions at IT companies even before graduation, and 46% of graduates occupied middle-level and higher positions at bigtech companies, major product companies, and startups.

A Bachelor’s program with Ozon in AI computer technologies

ITMO continues developing its collaborations with industrial partners. At the conference, ITMO and Ozon signed an agreement on the joint implementation of the top-level program Computer Technologies in AI within Bachelor’s programs Computer Technologies (01.03.02 Applied Mathematics and Computer Science) and Software Engineering (09.03.02 Information Systems and Technologies).

The program trains top-level AI specialists and is implemented within the federal project Artificial Intelligence and the national project Data Economics and Digital Transformation of the Country. Thanks to this collaboration with a major IT company, students will be able to access additional infrastructure opportunities, acquire relevant knowledge from industry experts, and solidify their skills on real-world business cases. According to Vera Manevich, the head of IT staff at Ozon Tech, the new Bachelor’s program is not the only collaboration format for Ozon and ITMO: together, the partners already implement the Master’s program Design and Development of AI Systems, host career days, and organize educational meetups for students.

“We are interested in attracting talented students and forming a talent pool. This means long-term investments into developing the ecosystem of domestic IT talents and in the company's technological leadership. Therefore, we actively engage in student training and participate in educational projects and initiatives. It’s important for us to be integrated into the educational and research processes in order to be able to form the future of engineering education together with universities; and ITMO is our key partner in this field. With educational projects, we’ll be able to contribute to the development of IT education, building a bridge between academia and big tech. This will allow the students to understand which knowledge and practical skills are important and valued on the market,” shared Vera Manevich.