Research
Research
Software engineering produces a great deal of opinion and a limited amount of evidence. This work is an attempt to change that ratio.
The research uses empirical methods, large-scale surveys, longitudinal studies, and structural equation modelling, to study how software gets built and what conditions make that work go well or badly. Every study is open-access. Data and replication packages are deposited in Zenodo for independent verification.
The work is organised into three pillars and a methodological base on which all three rest. Each connects directly to decisions that engineering teams and technology leaders face in practice.
AI in Software Engineering
AI is changing software development. Whether it is changing it in the ways the conversation assumes is an empirical question, and one this research takes seriously.
The core finding from adoption studies: workflow compatibility matters more than perceived usefulness. Engineers adopt AI tools that fit into how they already work. Tools that require a workflow change face persistent adoption resistance, regardless of how capable they appear on paper. The Copenhagen Manifesto translates findings from this programme into four operational principles for responsible AI integration: responsibility, transparency, inclusivity, and sustainability.
AI4SE1DK: Human-Centred Adoption of AI for Software Engineering in Denmark
A 48-month national consortium programme establishing Denmark as a global reference for trustworthy, human-centred AI4SE. Six universities, one GTS institute, and seven industry and public-sector organisations work together on three objectives: closing the adoption gap, enhancing technical trustworthiness, and reaching at least ten percent of the Danish software sector with concrete tools and practices.
People, Teams & Organisations
Software engineering is a human activity. The tools and methods matter, but so do the people using them and the conditions those teams work under.
Empirical studies across agile transformation, remote and hybrid work, and team diversity converge on a common thread: the interpersonal and organisational variables, psychological safety, social contact quality, management commitment, predict outcomes more reliably than technical choices. The research here puts numbers on that claim.
Education & the Next Engineer
How software engineers learn has received far less empirical attention than how they work. A body of studies covers the full educational pipeline: from high school students encountering computational thinking for the first time, through university-level work on cooperative learning and test-first programming, to the professional communities where practitioners exchange knowledge.
The consistent finding: active, participatory formats outperform passive ones at every level.
The Methodological Base
The three pillars rest on a shared commitment to rigorous empirical method. When controlled experiments are not feasible, which is often the case in software engineering, credible findings require principled alternatives. PLS-SEM, Empirical Standards, and emerging guidelines for LLM-assisted research are the analytical infrastructure that makes the findings in each pillar trustworthy.