Skip to main content

Building the National Data Factory and Testbed at AI Innovation of Sweden

Thursday, June 25, 2020

Technical infrastructure, data, legal frameworks, and know-how will be available for AI Innovation of Sweden’s projects and partners within its emerging Data Factory.

New competence supporting the Data Factory are project managers Ebba Josefson Lindqvist and Peter Bleckert as well as data scientist Sheetal Reddy.

"How can AI help my organization?"

Ebba Josefson Lindqvist, project manager, sees various needs in the partner community:

One of the main needs we see is knowledge about applied AI, we get questions about how AI can help organizations and what tools are available. We also see requests for understanding what the state-of-the art in the world is and how this knowledge can be accessed in different organizations.

Testing and experimentation is another area where the Data Factory supports the partners: 

"We see many situations where a partner organization wants to start using AI in their businesses, but lack sufficient computational power. To be able to try something out, they need a complete infrastructure adapted for AI projects."

An infrastructure catering for various needs cannot be a one size fits all. When it comes to hardware, the Data Factory will enable partners to work with both on-premise solutions as well as cloud solutions. Both solutions have their advantages, in some cases, regulations require on-premise solution physically located in Sweden for an idea to leave the concept stage and go into experimentation.

The current Data Factory infrastructure is donated and hosted in the Gothenburg-area by our core technology partner CGit. It comprises one NVIDIA DGX-1, with a one Petaflop performance, 40 000+ cores, and an appropriate storage capability for large projects and datasets.

Legal Expert group and Focus Group for Law

The datasets are equally important to get started on most AI journeys. Knowledge of legal frameworks, implications of technical solutions, as well as how to manage large and complex datasets, are high-demand resources in companies today. 

Several projects connected to the Data Factory have already been launched including the Road Data Lab, SCAPIS AI platform and the Swedish Medical Language Data Lab. In addition hackathons such as the Baltic Seabird Hackathon and Space Data hackathon have been hosted by AI Sweden.

The Data Factory functions as an access portal to large datasets, as a collaboration platform, as a compute resource, and as a knowledge repository.

"While we are continuously building on and expanding hardware and cloud solutions, we are also looking into and working to understand legal issues and develop use cases for how datasets can be applied and under various circumstances."

Two initiatives relating to legal issues are initiated; The Legal Expert Group and Focus Group Law.

We will extend an invitation to our partners to be part of a Legal Expert Group - a group comprising a smaller number of legal experts that will identify legal issues, discuss these and generate solutions and share their findings in for example, white papers or guidelines that could benefit all partners within AI Innovation of Sweden. 

Work within the second initiative, the Focus Group Law, has already started and here lawyers, researchers in law and others with an interest in legal questions are discussing these and suggesting initiatives or projects that AI Innovation of Sweden should run.

Future of the Data Factory

Data Factory is up and running but it will always remain a work in progress. Technology in this area is developing extremely fast and current solutions and knowledge will soon be outdated. This is the reason that this Data Factory is designated as a national testbed. The testbed capability will provide significant value to our partners and position AI Innovation of Sweden as an international benchmark of how to set up a highly valuable data factory.

Open for action

Today, partners of AI Innovation of Sweden can use the Data Factory to:

  • Donate datasets and models; use donated datasets and models
  • Develop and test models on datasets
  • Run pilot projects within the organization or together with other partners
  • Raise organizational know-how on how to set up and run a data factory of their own
  • Use compute power and infrastructure to solve challenges e.g. through hackathons
  • Use the Data Factory for research projects or master thesis student projects

If you are a partner and want to make use of any of the activities above, please reach out.

More about the Data Factory