To address the need for deliberation, decentralisation and democratisation in the next generation of cyber-physical systems, we converge the paradigms of self-organisation and multi-agent systems. This demands education in two dimensions for student engineers learning to design, implement and operationalise such systems. The first dimension is education in software engineering for a new style of system: dynamic, interactive, unpredictable, prone to error, self re-configurable, etc. The second dimension is education in ‘society engineering’, including elements of political science, psychology and economics. This covers an eclectic range of subjects including governance styles and structures; institutions and other social arrangements; deliberative assemblies and social choice; social influence and social networking; and collective action, civic engagement, and socially-constructed conceptual resources. This article reports on the design and delivery of an advanced, and inter-disciplinary, software engineering course in Self-Organising Multi-Agent Systems. This offers graduate computer scientists and electronic engineers a theoretical and practical grounding in both educational dimensions.