This team focuses on the fundamentals of machine learning and intelligence. For a long time, we have concentrated on language models and related topics, including interpretability, AI alignment, improved model architectures, multimodal LLMs, and reinforcement learning for LLMs. Our goal is to expand the community’s understanding of how to build better models and achieve stronger results.

Recommender systems are a key pillar of our research at T-Technologies, enabling personalized financial experiences. We focus on next-basket prediction, user interaction modeling, deep personalization, and robustness to data shifts for reliable recommendations in dynamic banking environments. We also explore RL for adaptive real-time personalization.

We architect next-generation algorithms and systems that redefine performance at scale. Our research bridges the gap between theoretical foundations and high-load production, creating breakthrough solutions for large-scale graph processing, NP-hard optimization, and distributed analytics engines.

Our goal is to boost engineering productivity by weaving AI throughout the software life cycle—from coding and testing to security and maintenance. Our research agenda goes beyond standard copilot tools to build autonomous multi-agent systems.