David Hästbacka studies AI-intensive software systems

For Associate Professor (tenure track) of Software Engineering David Hästbacka, artificial intelligence (AI) is a close collaborator. Hästbacka’s research focuses on software and system architectures. Based on these architectures, he develops smart software systems for applications that rely heavily on data and AI.

Hästbacka’s research has applications in domains such as industrial production environments, autonomous machines and energy systems.

“What all these domains have in common is a set of exacting requirements: the systems must be reliable, secure and resilient,” says Hästbacka.

Hästbacka’s research group is especially interested in how AI-intensive smart software systems should be designed and developed. One area of focus is the development of AI capabilities.

“As more data becomes available, AI models can be continuously refined and new features introduced,” he says.

Tampere University’s AI Cluster supports research on computationally intensive AI models

For David Hästbacka’s research group, access to a research infrastructure that is under their own control is essential. 

“When we are studying system solutions and assessing their performance, data latency and energy consumption, we aim to minimise the influence of external factors. When we control the platform and infrastructure ourselves, we can ensure that the results are not affected by variables beyond our control,” says Hästbacka.

Hästbacka has been involved in bringing Tampere University’s AI Cluster (AIC) under the umbrella of the Tampere Center for Scientific Computing (TCSC). The AI Cluster is available to researchers across the University and is particularly well suited to testing and running AI models. 

Hästbacka says that his group uses the AI Cluster especially for research that involves computationally intensive AI models. 

“For example, when we are working with sensitive data, this centrally managed service provided by the University complements public cloud services and the services provided by CSC, the IT Center for Science,” he says.

Hästbacka’s research group also has its own servers, edge computing devices, IoT equipment and communication networks, enabling them to study systems and platform solutions locally.

“Data is generated locally, and the results are increasingly utilised locally. We are exploring ways to also expand the local deployment of data processing and AI,” says Hästbacka.

Reliable data is critical in the age of AI

While the rapid advancement of AI is helping to drive research forward, it is also creating new challenges. According to Hästbacka, one of the key ethical questions in his field concerns the reliability of data and AI models. 

“Is our AI model based on reliable data? Does the data match the context in which we intend to use it? These are the kinds of questions that researchers should be asking.” 

Researchers must also consider technical practices, such as data access rights. 

Beyond the research community, the transparency of AI models has become a topic of widespread public debate.

“We need greater visibility into the data and materials on which these models are trained,” he says.

The risks associated with AI have received considerable public attention recently, and Hästbacka believes these concerns must be taken seriously.

“We need regulations governing the use of AI and data, but they must be designed in a way that does not unnecessarily hinder innovation or weaken Europe’s competitiveness,” he says.

AI has shifted the focus of software engineering 

The advancement of AI has rapidly transformed the field of software engineering. 

“In the past we only had to manage software releases, whereas now we also have to manage the data and AI models embedded within them,” says Hästbacka.

Hästbacka points out that software engineering has been at the forefront of the AI transformation from the outset. Many aspects of software development have changed significantly.

“There is less coding, as generative AI tools are increasingly taking over routine programming tasks. As a result, software development is moving towards higher levels of abstraction. Greater emphasis is now placed on specification, architecture and quality assurance, among other things,” says Hästbacka.

Watch a video showcasing David Hästbacka’s research.
Read more about the Software Engineering Research Center (TASE) at Tampere University.