Nikolay Donets of Revolut at RAAIS 2026

2025-10-09 · Source: Air Street Press · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cloud Computing & IT Infrastructure, FinTech & Digital Financial Services · Depth: Advanced, short

Summary

Nikolay Donets, Head of Machine Learning Engineering at Revolut, will speak at the RAAIS 2026 summit on June 12th in London. His organization builds the AI platform supporting all production AI systems at Revolut, which processes over \$1.3 trillion in transaction volumes and is a top finance app in 19 countries. This platform underpins diverse applications, from classical ML for fraud and personalization to time-series foundation models and voice agents. A notable implementation is the ElevenLabs-powered voice agent system, resolving customer tickets in under five minutes—8x faster—with a 99.7% call-handling success rate across over four million UK and European customers in 30+ languages. Donets emphasizes that the primary challenge in production AI is not isolated model building but creating a unified platform for builders, operators, researchers, and compliance within regulated financial products. He advocates for governance as a velocity enabler, proposing a 90-day framework for GenAI product launches under regulatory constraints, focusing on data lineage, continuous delivery, and compliance guardrails.

Key takeaway

For AI Architects and MLOps Engineers deploying AI in regulated sectors, prioritize building a unified platform that integrates governance, evaluation, and monitoring from the outset. Your focus should shift from isolated model performance to the surrounding infrastructure, treating compliance as a technical requirement. Adopt a framework like Revolut's 90-day GenAI launch plan, emphasizing data lineage and continuous delivery, to accelerate deployment while meeting supervisory demands. This approach ensures durable capability and avoids bottlenecks.

Key insights

The hard problem in production AI is building a unified platform for all stakeholders, not just isolated models.

Principles

• Governance enables velocity in regulated AI.
• Platform unifies builders, operators, researchers, compliance.
• Infrastructure around models is key to deployment.

Method

Nikolay Donets outlined a 90-day framework for GenAI product launches under regulatory constraints, based on data lineage, continuous delivery with multi-layered validation, and compliance guardrails.

In practice

• Implement multi-layered validation for continuous delivery.
• Treat compliance data as feature material.
• Integrate governance into development environment.

Topics

• AI Platform Engineering
• MLOps
• AI Governance
• Regulated AI
• Generative AI Deployment
• Financial Services AI

Best for: MLOps Engineer, AI Architect, Director of AI/ML

Related on AIssential

• Creating a new AI platform — Generative AI demands new platform capabilities beyond traditional ML infrastructure, even for established systems.
• Governing AI agents at scale with Unity Catalog — Effective AI agent governance requires controlling access and monitoring actions, not just reviewing potential behaviors.
• Snowflake and OpenAI: Transforming Enterprise Data Platforms — Embedding AI models directly within existing enterprise data platforms enhances security and governance for AI deployment.
• Podcast: 2025 Key Trends: AI Workflows, Architectural Complexity, Sociotechnical Systems & Platform Products — AI is rapidly reshaping software development workflows, amplifying existing organizational strengths and weaknesses.
• H2O MLOps Enterprise Model Registry & Hugging Face | Part 8 Integration | Part 8 — A robust model repository is essential for managing and governing a large, diverse portfolio of AI models in production.
• A Need for Nuance: The Economist’s Andrew Palmer — AI excels at rapid prototyping but faces significant challenges in production deployment and organizational integration.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Air Street Press.