Ingrid K. Glad

Integreat is recruiting 8 fully funded PhD fellows in Norway in knowledge-driven machine learning, with positions at the University of Oslo and UiT The Arctic University of Norway . One highlighted project is Data attribution for large language models , focused on understanding which parts of training data influence LLM outputs and how to disentangle factual versus linguistic effects. This PhD project sits at the intersection of computer science , machine learning , natural language processing , statistics , and mathematics . The research aims to develop scalable and informative attribution methods beyond simple scalar scores, with possible directions including Shapley-value approximations, influence functions, surrogate attribution models, and multi-faceted attribution frameworks. The methods will be evaluated in LLM settings such as fine-tuning, learning from examples, and open-weight language models. The position is affiliated with the Faculty of Mathematics and Natural Sciences in Oslo and is jointly supervised by Ingrid K. Glad and Martin Jullum . The broader Integreat call includes interdisciplinary projects spanning machine learning, statistics, logic, language technology, and ethics, and fellows join a collaborative research community with access to seminars, workshops, mentoring, and international collaboration opportunities. Eligibility highlights: applicants should hold a Master’s degree or equivalent in computer science (AI/ML/NLP), mathematics, statistics, or a related field, and have strong programming skills. Experience with NLP, LLMs, explainable AI, data attribution, Shapley values, and Python/deep learning frameworks is an advantage. Funding: the position is fully funded and lasts three years. Deadline: 9 August 2026, 23:59 CEST. Applicants submit a single application and rank up to three projects in order of preference.

Integreat and TRUST are recruiting PhD fellows in machine learning, NLP, and AI at the University of Oslo and UiT. One of the openings is Project #3: Data attribution for large language models , based at the University of Oslo, Faculty of Mathematics and Natural Sciences, Department of Mathematics. This project focuses on a major explainability challenge in modern AI: tracing which parts of the training data influence an LLM’s answers. The research aims to develop scalable data attribution methods that go beyond simple scores and can disentangle factual versus linguistic influences. Possible directions include Shapley-value approximations , influence functions , surrogate attribution models, and multi-faceted attribution frameworks. The methods will be tested in settings such as fine-tuning, learning from examples, and open-weight language models. Eligibility highlights: applicants should hold a Master’s degree or equivalent in computer science (AI/ML/NLP), mathematics, statistics, or a related field, and have strong programming skills. Experience with NLP, LLMs, explainable AI, data attribution, Python, and deep learning frameworks is an advantage. Funding: the position is fully funded and lasts three years. Fellows join a collaborative research environment with access to Integreat and TRUST activities, networks, seminars, workshops, mentoring, and interdisciplinary collaboration. Application window: the deadline is 9 August 2026 at 23:59 CEST. Applicants submit one application and rank up to three projects. An online information meeting is held on 22 July 2026 at 19:00 CEST.

Integreat is recruiting 8 PhD fellows in knowledge-driven machine learning in Oslo, Norway, with positions at the University of Oslo and UiT The Arctic University of Norway . The call is interdisciplinary and connects machine learning with statistics, mathematics, logic, language technology, ethics, philosophy , and related areas. The available projects include topics such as Bayesian inference and foundation models , logic, knowledge representation, and learning , data attribution for large language models , bias and fairness in vision-language models , machine reasoning , probabilistic representation learning , and structured reasoning with foundation models . Successful candidates will join the Integreat research community and also have opportunities to engage with TRUST – The Norwegian Centre for Trustworthy AI. Funding: These are full-time PhD fellow positions (paid employment). The post does not describe a scholarship programme; instead, fellows are employed by the host institution depending on the project. Eligibility: Applicants must meet the general qualification requirements, the PhD programme admission requirements, and the project-specific requirements for the ranked projects. A single application is submitted, and applicants must rank 1 to 3 projects based on their interests and qualifications. Deadline: 9 August 2026, 23:59 CEST . An online information meeting is scheduled for 22 July 2026 . How to apply: Read the project pages carefully, verify your eligibility, and submit one application ranking up to three projects. Use the official vacancy page for the full announcement and application instructions.