Nikolay Donets (Head of Machine Learning Engineering at Revolut) - RAAIS 2026

Revolut’s AI platform strategy centers on a centralized gateway architecture that streamlines governance and deployment across 40+ countries and 200+ products. By shifting the governance unit from individual models to specific AI use cases, the company effectively manages compliance, data sovereignty, and operational costs. A critical innovation includes a fallback chain for foundation models, ensuring system reliability even when primary vendors experience downtime. The platform’s evolution is exemplified by "Rita," an AI assistant that transitioned from a simple, rule-based chatbot to an advanced, multi-modal system capable of handling voice interactions and complex, real-time transactional data. This centralized approach reduces operational overhead and enables rapid scaling, allowing the company to maintain high performance while navigating diverse regulatory landscapes and evolving customer needs.