Skip to content
Portrait of Jens Dietrich

The Luvo 100

Jens Dietrich

Associate Professor of Computer Science
Victoria University of Wellington
New Zealand

Security & Assurance

Why Jens Dietrich is in the Luvo 100

Jens co-developed practical methods for comparing independently built software, including shared testing resources and the publicly available daleq4py tool. These resources support researchers and engineers investigating whether package differences affect behaviour or integrity. By explaining acceptable differences and identifying potential problems, the work strengthens checks on the reusable components that modern software depends on.

About

Jens Dietrich is Associate Professor of Computer Science at Victoria University of Wellington, working on techniques to check the integrity of software assembled from reusable components. Projects span Java and Python packages, tackling a practical question: when software is built independently from the same source code, do differences in the resulting files indicate a problem?

Jens co-authored Towards Cross-Build Differential Testing, studying 3,541 pairs of compiled Java packages. The approach runs the same generated tests against alternative builds and compares the results, looking for differences in behaviour that could signal compromised software or runtime failures. Publicly available results and reproduction tools allow other researchers to inspect and replicate the findings.

More recently, Jens co-developed daleq4py, a tool for comparing Python wheels using explainable rules. An accompanying study examined 12,180 releases from the Python Package Index, showing that many rebuilds can be accepted as equivalent despite differing byte for byte. Distinguishing harmless build variation from changes needing security investigation provides engineering teams with clearer evidence when assessing externally supplied software.

Career highlights

  • Co-authored Towards Cross-Build Differential Testing, cited at ICST 2025, studying 3,541 pairs of compiled Java packages
  • Shared cross-build testing results and reproduction tools on Zenodo in December 2024
  • Co-developed daleq4py to establish equivalence between independently built Python wheels
  • Co-authored No Snake Oil: Verifying Python Package Builds, evaluating 12,180 PyPI releases in 2026