Up to 5 PhD Fellows in Machine Learning at the Arctic Institute for AI in Science and Innovation

The positionThe PhD positions are based in [AI]² - The Arctic Institute for AI - at UiT The Arctic University of Norway.

[AI]², hosted by the Faculty of Science and Technology, is UiT’s new strategic flagship initative for interdiciplinary AI research and research-based innovation for meeting the major societal challenges of our time facing the Arctic, Europe and the World.
The positions will also be associated with the Visual Intelligence Research Centre and will be formally embedded in the UiT Machine Learning Group at the Department of Physics and Technology.

The positions are for a period of three years. The objective of the positions is to complete research training to the level of a doctoral degree. Admission to the PhD programme is a prerequisite for employment, and the programme period starts on the commencement of the position.

The workplace is at UiT in Tromsø. You must be able to start in the position within a reasonable time after receiving the offer.

Place of work[AI]² is a new strategic AI research initiative at UiT. Our mission is to contribute solutions to the societal challenges facing Europe and the world, particularly in the Arctic, a strategic frontier where environmental change, geopolitical competition, and emerging technologies are reshaping security dynamics. Important strategic focus areas for these positions are to develop solutions for increased preparedness and situational awareness leveraging satellites and UAVs, for climate and environmental forecasting and change, and for asset and infrastructure monitoring. We do this with “eallju” (Sami: eagerness to work) on a foundation of fundamental AI research in artificial neural networks and deep learning.

[AI]² is partially growing out of Visual Intelligence (VI), hosted by UiT as a Centre for Research-based Innovation (SFI) in AI funded for 8 years by the Research Council of Norway (grant no. 309439). The positions will be associated with VI to benefit from that research environment. In VI, the central topic is deep learning for industry innovation in the Earth, marine, energy and health sectors by developing novel methodology for learning from limited data, for uncertainty quantification, for incorporating context and dependencies, and for developing interpretable AI.

The UiT campus in Tromsø is located near the city centre. Tromsø is a vibrant city located in Northern Norway, with just shy of 80 000 inhabitants, surrounded by the stunning landscape of Northern Scandinavia. The location also offers ample opportunities for, e.g., sighting the northern lights, midnight sun, hiking and skiing.

The position's field of researchAs an [AI]² researcher, you will contribute to solving the pressing societal challenges of our time for a sustainable future. This means that the main aim is not to develop a favorite “AI hammer” for then to look for problems this tool may help solve. Rather, the aim is to identify particularly important problems to solve, for then to contribute to the development of particularly promising AI methodology suited for helping provide solutions for these real world problems and with real world stakeholders involved.

However, when developing the next generation AI technology, non-negotiable objectives are for the technology itself to be resilient, robust, and reliable, which are the hallmarks of responsible AI.

We are announcing up to 5 PhD positions. The intended thematic application areas for the positions are as follows:

1. 2-3 positions at the intersection of microclimate modeling and forecasting. Microclimate modeling refers to modeling of wind fields or other atmospheric or Earth surface variables at unprecedented temporal and spatial resolution. This is important for aviation and for autonomous operation of UAVs (unmanned aerial vehicles), e.g. in a search and rescue context. At least one position will be conducted in close collaboration with the aviation and drone group at UiT. Forecasting refers to prediction at short range, medium range or long range (e.g. decadal) of e.g., sea ice in the Arctic or other phenomena that depend on the changing weather and
Generert fra Jobbnorge.no tirsdag 27. mai 2025 kl. 16:00 Side 2 / 5
climate. It is also important for early warning systems, for instance, related to climate-induced extreme events, to help enable resilient societies for the future.
 

2. 1-2 positions at the intersection of situational awareness over sea and land and in the oceans by analysis of data from spaceborne satellites, UVAs, other observational measurements such as ocean echosounders or sensors at UUVs (unmanned underwater vehicles). Important challenges include, e.g., ship detection and monitoring, pollution and oil spill detection, and detection of situational anomalies in general.

3. 1 position related to infrastructure monitoring. AI-powered infrastructure monitoring helps organizations detect anomalies, predict failures, and resolve issues before they impact users or services.
 

For all positions, fundamental and methodological AI research underpins the development, with overarching objectives centered on:

Resilient AI: Develop AI technology that is resilient to changes and unforeseen events or attacks.Robust AI: AI that is resistant to variations, operating conditions, and drifts.Reliable AI: Interpretable AI for users, which is energy-efficient and aligned with fundamental human values.Potential underlying research directions to develop within these thematic areas and objectives are

Develop novel AI methodology which leverages numerical solvers (e,g for physical systems governed by partial differential equations) and which respects physical laws.Represent the general properties of relevant and real-world data by self-supervised learning towards AI foundation models.Represent general properties of data coming from different sources, i.e., multimodal AI models (combining images, text, etc). Understand the important mechanisms of the AI models in terms of their prototypical behaviour, individual neurons or layers within networks, or the quality of the data (data-centric intelligence).Develop interpretable generative AI solutions.Develop a new methodology to improve calibration and uncertainty quantification.Important

We expect all PhD positions to operate as a batch and to communicate and collaborate as a group. You must indicate in prioritized order the thematic application areas you are interested in.A detailed work plan and project description for the PhD candidate will be devised in collaboration between the fellow, the research team and the supervisors, as well as interdisciplinary collaborators.
Want to know more about the position?For enquiries about the position, please contact:

Director of [AI]² and Visual Intelligence, Professor Robert Jenssen: robert.jenssen@uit.noCo-Directors of [AI]² and Principal Investigators at Visual Intelligence
o Associate Professor Elisabeth Wetzer: elisabeth.wetzer@uit.no
o Professor Georgios Leontidis: georgios.leontidis@uit.noPrincipal Investigators
o Professor Michael Kampffmeyer: michael.kampffmeyer@uit.no
o Associate Professor Kristoffer Wickstrøm: kristoffer.k.wickstrom@uit.no
o Associate Professor Qing Liu: qing.liu@uit.noHead of Department of Physics and Technology, Professor Olav Gaute Hellesø: olav.gaute.helleso@uit.no.
Your role as a PhD FellowYou will be a part of [AI]². You will also be part of Visual Intelligence, and you will be formally embedded in the UiT Machine Learning Group. The [AI]² is in a build-up phase, but grows partially out of the Visual Intelligence Centre. For that reason, you will be part of the Visual Intelligence Graduate School (VIGS), a vibrant community of early career researchers within the centre. You will engage in collaborative research with the other members of [AI]² towards research-based innovation. You will invest considerable time and effort in interdisciplinary collaboration with researchers who are not machine learning experts but domain experts within the application field of your position.

You will be expected to actively collaborate outside of core machine learning with stakeholders, and for that reason, interest and experience with interdisciplinary research and innovation from a foundation of fundamental AI research will be considered positively. You are expected to contribute to [AI]² (including VI’s) virtual and physical seminars, be open to interdisciplinary collaboration and participate in data gathering and processing when needed. [AI]² and VI host the conference NLDL (Northern Lights Deep Learning Conference http://nldl.org) and you are expected to be involved in the organization.

QualificationsWe are particularly seeking candidates with a solid background in machine learning methodology, in terms of the mathematical and statistical foundation of such methods. We are seeking candidates with coursework and experience in deep learning, neural networks and machine learning, e.g. self-supervised learning, convolutional neural networks, transformer-based networks, statistical and information theoretic learning, geometric learning, Bayesian learning, (implicit) neural representations, neural operator learning.

Required qualifications:

You must hold a Master’s degree in machine learning or related fields within e.g., mathematics, statistics, computer science, physics, electrical engineering.A strong formal course background in deep learning and machine learning in general or relevant topics such as pattern recognition or computational statistics is required. Important topics are described above.Documented programming skills, for example, using Python, etc.Good communication skills in English are necessary and documented fluency in English is required. Nordic applicants can document their English capabilities by attaching their high school diploma.Preferred qualifications:

Research experience via a Master thesis, internships or similar involving the development o

This listing is from nav. View original listing ↗