Multilingual evaluation
Designing language samples and benchmarks, and testing whether findings transfer to underserved languages and different linguistic settings.
Typologically diverse evaluation · CL 2026 ↗Research collaboration
I lead AAU-NLP at Aalborg University in Copenhagen, Denmark, and welcome partners for Horizon Europe proposals and other European research consortia. I can contribute to or lead work packages on large language model (LLM) security and privacy, multilingual evaluation, and factual reliability.
Designing language samples and benchmarks, and testing whether findings transfer to underserved languages and different linguistic settings.
Typologically diverse evaluation · CL 2026 ↗Studying memorisation, embedding inversion and information leakage; investigating how linguistic properties shape attacks and defences.
Evaluating PII memorisation · ACL 2026 ↗Developing knowledge-grounded evaluation and methods for checking model outputs, including information extraction in specialist domains.
MultiHal · ICML 2026 ↗My security research is supported by the TRUST and (LM)²-SEC programmes. Recent applied collaborations include financial information extraction with ALIPES and educational feedback with LLM agents through SEFL in the Digital Twins for Abundant Feedback project. I also welcome joint doctoral supervision and research exchanges.
MSCA Postdoctoral Fellowships offer a route for researchers to develop a fellowship proposal with AAU as host.
Innovation Fund Denmark's Industrial Researcher programme supports joint research through Industrial PhD and Industrial Postdoc projects.
Discuss a proposal
Send a short outline, the call or topic if identified, the contribution you have in mind, and the deadline.