About the author

Artur Huk is a Cloud Data Architect at Future Processing. He also contributes to production-grade ML and LLM systems for the London insurance market—work where traceability, evaluation discipline, and compliance matter as much as model quality. His day-to-day sits at the intersection of data platforms, cloud engineering, and applied AI.

For roughly two decades he has delivered data architecture, analytics, and modernization programs—from enterprise data warehouses and lakes to cloud-native pipelines—often as the person translating business intent into durable engineering choices.

In recent years the focus has narrowed to Decision Intelligence, Responsibility-Oriented Agents (ROA), and the Decision Intelligence Runtime (DIR) patterns documented in this repository: explicit roles, boundaries, auditability, and deterministic execution around probabilistic models—so that agents are not treated as black-box orchestration loops.

The AIvestor R&D project was a hands-on stress test of those ideas: an event-driven agent architecture under real market constraints, with emphasis on explainability, decision history, and traceability rather than headline metrics. The project has concluded; the architectural lessons feed directly into the material you see here.

He is interested in calm, engineering-grade conversations about agents, architecture, and how organizations actually run when automation scales - LinkedIn.