Hannah Rohde

Two PhD positions are available in a project on Uncertainty Quantification and Communication for Large Language Models . The positions are based at the IT University of Copenhagen in Denmark and are jointly led by Christian Hardmeier and Jes Frellsen , with an extended research stay at the University of Edinburgh involving Hannah Rohde and Ivan Titov . The first position focuses on natural language processing and aims to develop methods that help LLMs express uncertainty in ways humans can understand, while aligning communicated confidence with the model’s internal state. The second position focuses on machine learning and aims to develop efficient methods for quantifying uncertainty in large language models. The project sits at the intersection of computer science , information technology , statistics , mathematics , and linguistics , with a strong emphasis on data science, experimental linguistics, semantics/pragmatics, and user studies with human participants. Applicants should have a relevant degree in computational linguistics, NLP, data science, AI, computer science, experimental linguistics, or a related field, plus strong Python skills. Experience with natural language data, PyTorch, LLM training/fine-tuning/evaluation, psycholinguistic studies, and academic publishing is desirable. These are fully funded PhD positions with employment and enrollment at ITU for 3 or 4 years depending on the applicant’s degree level. Salary and employment terms follow the Danish university agreement. The application must be submitted in English and include a cover letter, CV, degree documentation, statement of purpose, and English proficiency documentation, with publications or thesis recommended if available. Deadline: 24 June 2026 for the ITU position. The post also mentions a second related PhD position at DTU with deadline 7 August 2026.

The IT University of Copenhagen is advertising a fully funded PhD position in natural language processing and computer science focused on uncertainty quantification and how large language models can express confidence in ways users understand. The project combines NLP , machine learning , and experimental linguistics . The successful candidate will develop methods for controllable uncertainty expression in LLMs and study how communicated confidence aligns with users’ perceptions. The work includes both computational development and validation through user studies with human participants. The position is based at the IT University of Copenhagen and is jointly led by Christian Hardmeier (ITU), Jes Frellsen (Technical University of Denmark), and Hannah Rohde (University of Edinburgh). An extended research stay in Edinburgh is an essential part of the project. Eligibility depends on degree level: applicants for the 3-year route should hold a master’s degree or equivalent in computational linguistics, NLP, data science, AI, computer science, experimental linguistics, or a related field; applicants for the 4-year route should hold a bachelor’s degree in one of those areas plus one year of master’s-level study. Strong Python skills are required, and experience with natural language data, PyTorch, LLM training/fine-tuning/evaluation, semantics/pragmatics, psycholinguistic studies, and academic publishing are desirable. The application deadline is 24 June 2026 at 23:59 CEST. Applications must be submitted in English through the university’s e-recruitment system and include a cover letter, CV, degree certificates, statement of purpose, and proof of English proficiency. Publications or a master’s thesis are recommended but not required.