Skip to content

About

I am an Associate Professor of Computer Science at the University of Iceland, where since 2026 I have also been Head of the Faculty of Industrial Engineering, Mechanical Engineering and Computer Science. I am a Principal Scientist at Amgen deCODE and a co-founder of KatlaCode. My group works on natural-language processing for low-resource Germanic languages, primarily Icelandic and Faroese. We also work on computer vision for the natural sciences (for example, ageing fish from their otoliths with the Marine and Freshwater Research Institute), clinical AI (automated coding of medical records with Landspítali) and human genetics (genome-wide studies of common diseases at deCODE).

I trained at ETH Zurich (PhD 2017) in mathematical models of neural systems. After a postdoc in systems neuroscience I moved home to Iceland and spent two years as a data scientist at Íslandsbanki. In 2020 I joined the University of Iceland, where I have built a group around language models and datasets for under-resourced languages. The group currently has three PhD students. I co-founded Prescriby (2019–2023) and, in 2024, the University of Iceland Centre for Artificial Intelligence (Gervigreindarsetrið).

Talking about AI

A growing part of my work happens outside teaching and research. I use AI every day and am regularly startled by how fast it moves, so I write op-eds, give talks to anyone from nurses to central bankers, and turn up on radio and television to discuss what it means for Iceland. A list of interviews and articles is on the outreach page.

In October 2026 I made Could AI destroy us? (Gæti gervigreindin grandað okkur?), seven short films that each retell one serious argument about how AI could go badly wrong. They are on YouTube in Icelandic and English, and on TikTok. I also wrote a post about how I made them.

Beyond research

I came to research through an interest in mathematics and computer science. That led me into neuroscience, where I asked how small circuits of neuron-like units add up to the computing power of the cortex. I don't expect to find the "canonical microcircuit" any time soon, but the mathematical grounding and the scientific habits I picked up in those early years stay with me, and they shape how I work on language and AI today. There's an Icelandic saying for it: lengi býr að fyrstu gerð, what is first shaped lasts long.

CV

Short CV (PDF) · Full CV with all publications (PDF) · View inline

For journalists

I am happy to talk to the media about AI and society, large language models, AI in education, AI risk and safety, and language technology for Icelandic. Email is the quickest way to reach me: hafsteinne@hi.is. Earlier interviews and articles are on the outreach page.

Short bio

Hafsteinn Einarsson is an associate professor of computer science and Head of the Faculty of Industrial Engineering, Mechanical Engineering and Computer Science at the University of Iceland, and a principal scientist at Amgen deCODE. His research covers language technology for Icelandic, the evaluation and safety of AI systems, and machine learning in science and medicine.

In Icelandic

Hafsteinn Einarsson er dósent í tölvunarfræði, deildarforseti Iðnaðarverkfræði-, vélaverkfræði- og tölvunarfræðideildar Háskóla Íslands og sérfræðingur hjá Íslenskri erfðagreiningu. Rannsóknir hans snúa að máltækni fyrir íslensku, mati og öryggi gervigreindarkerfa og vélnámi í vísindum og læknisfræði.

Download a photo (960 × 1280) (free to use alongside coverage of my work).

Contact