AI-enabled systems increasingly influence high-stakes decisions, yet their opacity and unclear accountability mechanisms hinder their safe and trustworthy deployment. The state of the art lacks a proper software engineering oversight mechanism to audit AI systems (internally and externally) to achieve compliant-by-design AI systems prior to deployment. This research investigates how to enable verifiable and auditable AI systems by design, addressing gaps in intellectual-property-preserving transparency, operationalisation of regulatory requirements, and lifecycle-wide auditing. The research’s expected outcome is a unified framework and accompanying tools to support compliant-by-design, trustworthy AI systems across their development and operational lifecycles. All contributions will be empirically evaluated through case studies and interviews.