Speakers
Marian Faryna, Head of Data Engineering CoE, Sigma Software
This talk explores how to turn EU AI Act requirements into a concrete, owned architecture backlog, rather than treating compliance as a separate activity performed after implementation. It covers key areas such as data residency and accessibility, logging and observability, dataset registration and lineage, data portability, residency mapping, and auditability.
The objective is to establish compliance by design, embedding the required governance, security, and traceability capabilities directly into the data and AI architecture. This provides a robust foundation for future AI use cases and reduces the associated risk, costs, and rework involved in retrofitting compliance requirements during migration or later stages of platform evolution.
