Scientists who require advanced software and high-performance hardware for their research were, until recently, largely dependent either on their own development or on collaboration with technically oriented departments. Today, the integration of research with cutting-edge information technologies — including artificial intelligence — at Charles University is ensured by a new specialised Research Software Engineering (RSE) Centre, the first of its kind at Czech universities.
A team led by Dr Zdeněk Mašín from the Department of Theoretical Physics at the Faculty of Mathematics and Physics offers services across the university and is already involved in a number of projects. “An interdisciplinary approach helps to find original solutions to complex problems, and modern AI tools open up new possibilities for this purpose,” says Mašín. The centre therefore focuses, among other things, on the development of agent-based environments linking domain expertise, databases, and computational infrastructure.
Reducing Barriers for Researchers
Are you a radiologist or an orthopaedist needing to analyse large volumes of ultrasound images? An RSE specialist will help you formulate research questions and propose an appropriate software solution — or build a custom application. This application can run either on your laptop or on the university’s Chimera computing cluster and can be operated via an AI assistant, a web interface, or the command line. The result is less time spent on technical tasks and more time for research itself.
Large-scale analyses cannot be carried out without powerful hardware. “For example, in sequencing the human genome, it is not so much about computational power as it is about handling tens of terabytes of data, which is enabled by the university cluster,” explains Dr Jan Stuchlý from the Second Faculty of Medicine. In this environment, AI agents can store, analyse, and efficiently combine data.
A special tool being developed by the centre in cooperation with the Central Library is the AI Sandbox — an environment where local large language models run without the need for cloud services, enabling researchers to experiment with AI securely. Crucially, sensitive data remains within the university infrastructure.
Origins of the Idea
The data centre in Prague-Troja, where the RSE is based and where the computing clusters are located, was completed in 2026. However, the history of data centres — now standard in research organisations and universities in Western Europe — dates back to the mid-20th century. The gradual development of various scientific software tools led to the need to incorporate experts working at the intersection of research and IT into the institutional framework of science. Research software engineering, in its modern sense, has been discussed since the 2010s and is far from limited to data centre computations. It encompasses a wide range of software supporting research across disciplines — from the humanities to the natural sciences.
“I first heard about research software engineering at a conference in Oxford in 2012 — coincidentally near the university where I studied. When I returned to the Czech Republic in 2018, after a stay in Germany, I quickly recognised the significant development potential in this field,” recalls Zdeněk Mašín.
A Flexible Operating Model
For the Czech version of the data centre, Zdeněk Mašín designed an organisational structure tailored to Charles University:
“Unlike Heidelberg or British universities, where RSE teams typically operate independently and are not directly part of individual faculties, we have adopted a hybrid model. Software engineers are centrally coordinated and work as a team, but they are employed directly by specific faculties to remain close to their scientific domains.”
The Czech centre has another unique feature. “At many foreign universities, software development teams — RSE — and high-performance computing (HPC) teams are separate. Our model is somewhat unusual internationally in that we offer both expertise and hardware infrastructure as a unified whole. This enables much faster communication and deployment of projects — for example, AI agents — which would take significantly longer in the traditional model due to more complex negotiations,” adds Dr Stuchlý.
Who Is Involved
The project currently involves staff from the Faculty of Mathematics and Physics, the Central Library, the Faculty of Arts, and the Second Faculty of Medicine. “When Dr Mašín approached us in 2022, I immediately saw a huge opportunity for the university. We have been successfully supporting data analysis and processing at our faculty for years, but a university-wide RSE team and the power of the HPC cluster in Troja significantly expand our capabilities,” says Dr Karel Fišer, Head of the Institute of Bioinformatics.
The centre is open to all disciplines. In addition to processing scientific data and complex computations required by natural science and medical faculties, it can assist with analysing large text corpora — for example, in creating a web-based Egyptological database, typical of humanities and theological faculties. Projects leveraging AI and large language models — especially in medicine — are rapidly gaining importance.
A Unique Initiative in the European Context
The initiative by Charles University researchers is exceptional, among other reasons, because in Western and Northern Europe, scientific data centres are typically overseen by national associations and directly supported by governments. The Charles University initiative is progressive even on a national scale, particularly in the context of rapid AI development. The European Union has made the construction of several strategic AI “gigafactories” a key tool for strengthening competitiveness and the digital economy, and the Czech Republic is interested in participating.
From Research Infrastructure to the Philosophy of Knowledge
The RSE team sees rapidly developing AI agent technology as a major opportunity for the university. It has the potential to fundamentally transform interdisciplinary research and reshape how we perceive and organise knowledge. The development of AI agents and increased access to data centres will not only make scientists’ work easier but could also change how we understand knowledge itself.
“Unlike libraries, which are repositories of knowledge, agent systems will have the ability to combine knowledge in ways that are difficult for domain experts to achieve. We see this as a major opportunity for future research. Our knowledge — like books — is typically structured by discipline, but connecting fields such as biomedicine and physics requires more than domain expertise. We want to explore how to define and handle these unique capabilities. This includes, for example, codifying relationships between concepts and knowledge (knowledge graphs), as well as specific experiences and rare or edge cases that often exist only in experts’ minds. Yet these are precisely the cases that have led to groundbreaking discoveries, such as CRISPR technology or pulsars in physics. Are these types of information essential for enabling agent-based systems to creatively solve problems across disciplinary boundaries?” reflects Zdeněk Mašín.
