Insights from a Student-Postdoc Tandem for Research Data Management in an Interdisciplinary Collaborative Research Center

Collaborative research projects often employ research data management (RDM) practices, such as data sharing and archiving strategies, IT services, and training to implement the FAIR (Findable, Accessible, Interoperable, Reusable) data principles [1,2]. These practices are typically embedded by Data Stewards with scientific or technical expertise [3]. However, routines for FAIR handling of research software [4] are not broadly established. In the Collaborative Research Center (CRC) 1294 – "Data Assimilation," where research focuses on theoretical mathematics and empirical sciences, the employed RDM strategies aim to enhance reproducibility, collaborative work, and compliance with German Research Foundation (DFG) regulations. One particular strategy involves integrating code review and consulting through a tandem consisting of a data steward (researcher with expertise in empirical sciences) and packaging experts (student assistants in data science). Here, the data steward conducts training and community engagement, while packaging experts focus on data and code verification to ensure reproducible research. In this contribution, we share our experiences from this collaboration. We observe that implementing RDM practices increases the reproducibility of scientific results [5]. However, their successful integration depends – besides structural support – on supervisor commitment, individual motivation, and research culture. For example, while data and software often play a central role in empirical sciences, research methodologies in theoretical sciences frequently diverge from traditional data- and software-focused RDM practices, leading to uncertainties about community-specific needs [6]. Furthermore, RDM practices are sometimes perceived as a service task to be delegated rather than an inherent scientific duty, highlighting the need for consistent education about the goals of RDM. Over time, success stories, like discovering code errors, help highlight the value of RDM practices. Nevertheless, some scientists remain unsure how to integrate existing support or services, pointing to structural communication gaps. Regarding IT infrastructure, we use external and self-hosted solutions but encourage using institutional platforms for collaborative coding, writing, and storage. These platforms are generally well-accepted, but some researchers underutilize these resources or encounter administrative barriers to interinstitutional collaborations. This limitation also affects the CRC-internal archive hosted to meet DFG guidelines. The archive's operation requires constant IT maintenance, coordination with the university's central IT, and occasional workarounds for external collaborators. Thus, researchers call for bridges towards the NFDI, but these require active commitment from the disciplines to be effective. According to our observations, we advocate for stronger integration of RDM staff in interdisciplinary research projects and active engagement with national RDM initiatives. In theoretical research, RDM dialogues must intensify to increase the visibility of the fields' needs and ensure verifiability and knowledge transfer. Whenever possible, researchers should rely on institutional IT services to ensure long-term maintainability and use Git and containerization strategies to improve the long-term use of empirical scientific methods. At the same time, clearly defined tasks, goals, and responsibilities help RDM team members understand their roles faster and support smoother integration into projects, as researchers highly appreciate code review and consulting. On the national level, interdisciplinary RDM initiatives are crucial in harmonizing and advancing RDM practices and services beyond institutional boundaries and discipline-specific research cultures.