T-Technologies Group on open source in AI/ML: how LLM work is organized inside the company

T-Technologies Group has openly described how internal AI/ML work is organized at one of Russia's largest technology organizations. Two key leaders of the division gave the interview: Anatoly Potapov, who heads the foundational LLM technologies group, and Daniil Gavrilov, who leads artificial intelligence research.

T-Technologies Group is the technology core of the T-Bank ecosystem. In recent years the company has been actively building up its AI/ML expertise, from applied models for banking products to fundamental research on large language models. The team's task is not simply to deploy ready-made solutions, but to develop its own technologies and take part in shaping the open source ecosystem around them.

Anatoly Potapov oversees the division responsible for the basic components of the LLM stack: architectures, training methods and infrastructure for working with large models. According to him, betting on open source code not only speeds up the company's own development but also helps attract strong engineers, those who care that their work is visible and reproducible outside the company.

Daniil Gavrilov is responsible for the research side: experiments, publications and the search for new approaches. He describes the team's structure as a hybrid of an academic lab and a product department, with an emphasis on making research have practical output rather than remain only on paper.

Both leaders agree on one point: openness is not a PR move but a working tool. When internal work is published openly, it creates external feedback, helps catch errors earlier and builds a reputation that converts into hiring.

This format, in which a large Russian company publicly walks through the inner workings of its AI development, is still rare. The interview with the T-Technologies team is interesting not only as an industry case but also as a signal: Russia's corporate AI sector is developing its own culture of openness, even if a cautious one.