This year, the open call consisted of three tracks. New Frontier Laboratories With a Broad Focus, the first track, was targeted at teams ready to work on breakthrough research in various fields. The main condition was that the applicants’ chosen topic had to align with the global scientific trends. Moreover, the application had to be supported by a researcher from the organization that was to be the project’s strategic partner.

The track New Frontier Laboratories in AI was open to research teams with ideas for interdisciplinary solutions in AI for specific fields – physics, chemistry, biology, and others. Finally, the track AI in Education accepted applications for projects aimed at forming an environment for a conscious and responsible use of AI service in education. 

According to Alexey Slobozhanyuk, the scientific consultant of the first two tracks, this logic falls within the main objectives outlined by the federal program Priority 2030.

“Following the guidelines of Priority 2030, ITMO develops frontier science that will become a part of our university’s foundation. That’s why it’s important for us to support new frontier laboratories in various fields. This time, we’ve also created a track focused on AI and AI-based solutions for specific problems in adjacent fields. We expect that the laboratories with a broad focus will help the university establish cooperation with key academic partners around the world, while AI-focused laboratories will promote interdepartmental cooperation and collaboration with prominent tech companies such as Sberbank and AIRI,” explains Dr. Slobozhanyuk. 

ITMO staff and students, as well as representatives of other Russian or international institutions, could submit their applications. For this, they had to choose a track and submit their applications on the open call’s official website. All applications passed a two-step evaluation. First, they would need to meet the competition’s criteria. In the second and main stage, the projects were reviewed by leading experts from ITMO and other Russian and international universities with experience in the field of the frontier laboratory. They took into account each project’s relevance and significance, as well as its place on the global scientific frontier. Applications in the track New Frontier Laboratories in AI were also reviewed by ITMO’s Council for AI.

In the final stage, participants attended open project defenses, wherein they had to explain how their projects solve a relevant problem, explain the benefits of their solution, and describe the result and its indicators, as well as the future prospects of their projects. 

11 projects were named winners as a result of all three selection stages; seven projects within the track New Frontier Laboratories With a Broad Focus and four projects within the track New Frontier Laboratories in AI. You can find the full list of winning projects here (in Russian).

A total of 41 applications were submitted during this open call, with the greatest number of projects (21) falling within the track New Frontier Laboratories With a Broad Focus. 

Winners will receive access to the expertise, staff, and infrastructure of ITMO and its partners, as well as financial support (up to 20 million rubles per year in the track New Frontier Laboratories in AI and up to 10 million rubles per year in the track New Frontier Laboratories With a Broad Focus).

All the winning projects will receive funding starting from 2025. In the first year, the teams will be fully supported by ITMO; next, the teams will have to attract funding on their own (no less than 50% in the second year and 100% – in the third year). Moreover, the teams will be presenting the results of their work twice a year. 

Winners 

New Frontier Laboratories With a Broad Focus

• Development of Advanced Flexible Optical Sensors for Monitoring of Health Indicators

This project, headed by Sergey Makarov, the chief researcher at ITMO’s Faculty of Physics, was considered among the best ones in its track. Its key idea is to develop flexible optical sensors based on new nanomaterials with improved spectral characteristics. Such sensors can simultaneously measure several indicators, such as pulse, optical cardiogram, oxygen saturation (SpO2), and glucose levels.

“There is currently a high demand for flexible wearable devices for health monitoring that can be conveniently attached to the human body. However, such devices are often expensive while lacking reliability. In collaboration with medical institutions, we are developing prototypes of non-invasive, multichannel flexible devices for monitoring several health indicators such as pulse, optical cardiogram, oxygen saturation, and glucose levels. The advantage of our devices lies in the narrower spectral lines of the sensor's light-emitting unit, which makes it possible to get a more selective response from biomolecules. With this feature, it’s possible to reduce cross-absorption from the main chromophores (biomolecules that absorb light and interfere with measurements) in the skin and enable the creation of a multichannel sensor for simultaneous detection of oxygenation, pulse, and sugar levels. Additionally, the suggested platform will be aimed at creating patches that can be attached to any area of the skin,” explain Sergey Makarov and Maria Sandzhieva, a junior researcher at the Faculty of Physics.

Among other noted projects in the track were projects for the creation of laboratories for studies of axions, which are potential dark matter particles; as well as the creation of nanophotonic metastructures for ultrafast calculations and semiconductor heterostructures for information transfer and processing in optoelectronic devices. 

New Frontier Laboratories in AI

• Smart Assistant for the Development of New Treatments and Materials

This project by Anna Kalyuzhnaya, a senior researcher at ITMO’s Research Center “Strong AI in Industry,” aims to develop an LLM-based smart digital assistant for chemists. Under its hood are LLMs that will manage chemists and ML agents, helping them search for information, create new molecules, predict their properties, and plan experiments. All of this will facilitate the creation of new treatments and materials for staff of research laboratories, pharmaceutical companies, and chemical companies. Additionally, the digital assistant’s core will be adaptable for other fields; moreover, chemists won’t have to learn to code to be able to use it.

The project will be implemented in collaboration with Sberbank. In three years, the team is planning to go from proof of concept to a full-scale product, solving any tasks that arise along the way. First, the digital assistant will learn to work with information sources online; then, the researchers will improve the program’s capabilities in generating molecules and materials and predicting their properties; and finally, all these properties will be united and the model will be tested on real-life cases with partners.

• Platform for Virtual End-to-End Screening of Candidate Chemical Compounds via Multimodal Models

Another winner in the track is Nikita Serov, the head of the Laboratory for Generative Design of Enzymes and Aptamers. With his colleagues from ITMO’s Center for AI in Chemistry, Dr. Serov suggested the development of an open-code platform for end-to-end screening of candidate compounds. With the platform, medical chemists, biotechnologists, and materials scientists will be able to predict complex properties of chemical systems such as their interaction with target proteins, catalytic activity, or behavior in living organisms. In turn, this will help them create antibacterial treatments and inorganic catalysts, which will, in fact, feature in /the platform's first practical testing.

Powering the platform will be an ensemble of pretrained multimodal models; moreover, the platform will include modules for the collection, augmentation, and preprocessing of chemical data, model-building, and validation of candidates with theoretical modelling methods; also featured will be a general virtual screening pipeline for several major classes of chemical systems: small molecules, polymers, and crystalline materials.

• New Modeling Methods for Studying Active Matter

Developed by Nikita Olekhno, a researcher at ITMO’s Faculty of Physics, and Ilya Makarov, an associate professor at the Institute of Applied Computer Science, this project is aimed at studying accumulations of particles, each of which is capable of transforming its inner energy into directed motion. Among such particles are both natural objects: groups of animals, bacteria clusters, and organic tissue; and artificial ones – from robot clusters to colloidal particle solutions.

As part of the project, the team will develop new methods for modeling such environments, as well as new ways to control their self-organization using machine learning and AI. This approach should prove more efficient than existing solutions by leveraging automated execution of a large number of experiments with clusters of dozens and hundreds of small moving robots and using this data to train AI systems.

The project’s results will advance the fundamental physics of such ubiquitous environments and help develop applied solutions based on controlled dynamic restructuring of matter at the macro or micro level – in robotics, microfluidics, and colloidal systems. It’s planned to develop the lab in collaboration with researchers from AIRI, with a strong focus on creating open-source software based on the obtained results.

• AI Models for the Design of Complex Engineering Solutions

In this project by Alexander Khvatov, the head of the Laboratory for Composite AI, the researchers will be training a fundamental AI model featuring open-source code, weights, and architecture, that will significantly accelerate the design of complex engineering solutions. Currently, it takes considerable time to develop one complex device with the help of precise physical equations. Instead of producing texts, this fundamental model will be able to predict the physical frameworks for the selection of geometric and other parameters of the new device with the help of a simplified model built on a physical core. This will significantly facilitate the design of neuromorphic computers – which make working with neural networks faster and cheaper – as well as engines, lasers, and microfluidic devices.