Barth W1, Smithies J1, Bettinson M1, Lei J1, Sefton P1
1Australian National University, Acton, Australia
Biography:
Wolfgang Barth is a Research Software Engineer bridging research communities and technical development at ANU. Working across language documentation and data infrastructure projects, he coordinates between Indigenous communities, data stewards, and IT teams to design solutions that balance technical excellence with ethical data governance aligned with FAIR and CARE principles. In the RSE Capacity Enhancement Project, Wolfgang works to strengthen research software engineering capabilities across the HASS and Indigenous research sector by developing accessible guidance, standards, and patterns that support both professional developers and researchers building computational tools.
Abstract:
The Humanities, Arts, Social Sciences, and Indigenous (HASS & I) research community faces a persistent shortage of Research Software Engineering (RSE) expertise. This is compounded by fragmented practices and limited access to sector-specific guidance. As AI integration becomes an urgent practical concern, the need for principled, community-owned technical resources has become critical.
The Research Software Engineering Capacity Enhancement Project (RSE-CEP), funded through the ARDC HASS and Indigenous Research Data Commons Community Data Lab, is building three interrelated outputs: a recommended patterns guide, architectural principles, and a technology roadmap. This presentation focuses on the processes that build and the structure that forms the pattern guide.
A public co-design workshop in March 2025 validated the problem space and shaped priorities. From this foundation, the project has developed a human-centred authoring workflow: patterns are written by team members and external contributors, using a structured template of frontmatter metadata and prose description grounded in general principles. Drafts are peer reviewed in GitHub before publication. An AI-assisted quality and conformity check runs prior to publishing to link the related patterns and to support consistency across the growing collection. Patterns are organised as a pattern language, with clusters of related patterns. They are published to a static website for community use and contribution.
This presentation reflects on the design decisions behind this approach: why human authorship and peer review remain central, how AI is used narrowly for structural consistency rather than content generation, and what sustainable community-extensible RSE infrastructure looks like for HASS practice. This abstract has been spell-checked using Claude Haiku.
