On July 18, T-Bank will hold its third Turbo ML Conf at the Serp i Molot cultural center in Moscow, assembling over 700 ML engineers and AI researchers from Russian tech companies. The conference targets practitioners who push machine learning models into production, not just theoreticians.

Dmitry Ushanov, director of T-Bank’s AI center, confirmed the core tracks remain: large language model fine-tuning, AI-assisted development, and multimodality. This year’s program splits into three parallel tracks covering architecture and training of modern models, interpretability, safe behavior, and reasoning with self-correction; product integration of ML, blending classical and AI models, ensuring predictability, and measuring impact on user experience and business metrics; and the technical core of ML systems—low-level infrastructure, fine-tuning methods, and system architecture.

More than 20 speakers from T-Bank, Yandex, Avito, Sber, and other firms will present case studies and technical deep dives on building scalable solutions, integrating specialized AI assistants and visual models, and recent research on LLM training. A demo zone will showcase product and platform solutions leveraging ML.

Attendance is free but registration is required and space is limited. For data and ML engineers, this is a rare chance to compare architectural trade-offs and infrastructure patterns across leading Russian tech stacks without the usual vendor pitch noise.

Source: RIA Novosti