Impact

Real projects on our infrastructure

The projects, outputs and events built on DHinfra — and how to credit the infrastructure in your work.

Projects building on DHinfra

For now this lists projects using the GPU & LLM cluster. A curated subset; each row links to the project website, a blog entry where available, and the funding body.

ProjectLinks
Historical Job AdsWebsite · FWF P35783
Austrian Newspapers Online (ANNO)Website
Unlocking the SchematismusFWF FG3100
From Digital to Distant Diplomatics (DiDip)Website · EU 101019327
Managing Maximilian (ManMax)Website · FWF F92
Aldersbach Abbey account booksWebsite · Blog entry · FWF/DFG PIN2678923
Medieval Serbian Charters Action (MeSeCa)Blog entry · EU 101154457
German Arithmetical Treatises in Manuscripts of the Late Middle Ages (ARITHMETIC)Website · EU 101039572
Celtic and Latin Glossing Traditions (GLOSSIT)EU 101123203
Moving in Space and Time (MoST)FWF PAT1763723
Tartinians in EuropeWebsite · FWF/DFG/SNSF PIN5035023
Autor:innenbibliotheken als Wissensräume: Digitalisierung des Katalogs der Arnim-BibliothekCLARIAH-AT

Outputs

Publications

  • HPC and DHinfra published
    Atzenhofer-Baumgartner. ASHPC24 (Austrian–Slovenian HPC Meeting), Grundlsee. 2024.
  • Design and Operation of a Federated GPU Cluster for Digital Humanities within DHinfra.at published
    Atzenhofer-Baumgartner, Fleischhacker, Resch, Waldhofer, Otto. ASHPC26 (Austrian–Slovenian HPC Meeting), Vienna. April 2026.
  • Line Extraction on Medieval Charters with Mask R-CNN accepted for publication
    Renet, Nicolaou, Vogeler. HIP'26 — 8th International Workshop on Historical Document Imaging and Processing, at ICDAR 2026, Vienna · LNCS, Springer. September 2026.
  • Design and Operation of a Federated GPU Cluster for Digital Humanities within DHinfra.at — extended version in preparation
    Atzenhofer-Baumgartner, Fleischhacker, Resch, Waldhofer, Otto. in preparation (arXiv).
  • Evaluation and Funding of Free and Open Source Software for Digital Humanities within DHinfra.at in preparation
    Atzenhofer-Baumgartner, Heine, Trognitz, Wachter, Waldhofer, Scholger, Schlögl, Ďurčo, Vogeler, Wallnig. in preparation.
  • Open-source tooling for video retrieval with Vision Language Models
    Ferdinand Köppen. University of Klagenfurt (AAU), master's thesis.

Events

Workshops and university teaching that made use of the GPU & LLM cluster.

Workshops

Teaching

CourseInstitutionTerm
Course Project in Artificial Intelligence and CybersecurityUniversity of Klagenfurt (AAU)Winter semester 2025/26
Projektseminar (project seminar)University of GrazSummer semester 2026

Giving credit

If the infrastructure helped your work, please acknowledge it. Whatever sentence you use, keep one string intact — the full name with the abbreviation: “Digital Humanities Infrastructure Austria (DHinfra)”.

Depending on what you used, for example:

UseSuggested wording
GPU & LLM clusterThe computational results have been achieved using the Digital Humanities Infrastructure Austria (DHinfra).
Digitization & imagingThe digitization was carried out using the Digital Humanities Infrastructure Austria (DHinfra).
SaaS / IaaS & storageThis work was supported by services of the Digital Humanities Infrastructure Austria (DHinfra).
Mixed or unsureThis work was carried out using the Digital Humanities Infrastructure Austria (DHinfra).

If the infrastructure covered only part of the work, add “in part” — e.g. “achieved in part using the Digital Humanities Infrastructure Austria (DHinfra)”.

This is not just for papers: please add the same acknowledgement to datasets, models and code you publish (for example on Zenodo or Hugging Face).

Citing the GPU & LLM cluster

If you used the GPU systems, we ask that you additionally cite the cluster. The paper describing the design and operation of the federated GPU cluster (presented at ASHPC26, Vienna, April 2026) is being prepared for arXiv; a citable reference with a DOI will be linked here as soon as it is available.

Collaboration & support

When we collaborate, it often makes sense to budget for HPC / MLOps support within a project. In our own work, and as we see at other universities and in industry, this is an efficient way to accelerate research. We frequently provide code support anyway, so some form of collaboration usually exists already.

Where contributions are substantial, this can also be arranged so that the admins involved are credited as co-authors on the resulting output. That keeps our work visible — which matters both to us and to funding agencies. Where someone from DHinfra joins the author team, a contribution taxonomy such as CRediT is a simple way to make each person's role explicit.

Want to be featured — or did we miss something?

If you use DHinfra and would like to appear here as a project, use case or output, or if we have overlooked your work, please let us know. You can open a pull request adding an entry to our repository, or simply write to us.