Centria: Generative AI in education and research

Centria AMK: University of Applied Sciences

Centria University of Applied Sciences is a major educational institution in Finland with extensive research and development operations. They have around 300 staff including employees and teachers. More than half of them are involved in R&D implementing projects  with organizations across various sectors.

Challenges They Faced

Centria's R&D department needed a platform for developing AI projects. They had a need for both, managing internal projects and providing solutions for customers. It was a challenging task because it required a secure, closed-loop system for handling sensitive data while supporting diverse project requirements. Since they work with many different sectors, the solution had to scale across various use cases. For example, they work with companies that generate time-series data, so the system needed to handle this type of data as well as any other data they might encounter in the future.

Why They Chose ConfidentialMind

Centria was excited to learn about our generative AI infrastructure platform capabilities, especially since it was designed for on-premises, closed-loop systems. Their main use for our platform was to integrate it with their infrastructure to develop generative AI applications. The database deployments and user authentication were seen as benefits. They recognized that their customers were constantly asking how to run AI systems in their own infrastructure, making ConfidentialMind Platform particularly appealing.

What the ConfidentialMind Platform Solves?

ConfidentialMind platform addresses Centria's key requirements by providing a comprehensive stack where they can run any containerized workload on Kubernetes cluster. The system supports multiple teams, which was one of the requirements. Also, the platform offers flexibility and resource optimization. With features like GPU limits with the current infrastructure and the ability to 'stop' running a model to free up space and schedule time on the environment however they wish. The solution also caters to their need for different open-source LLM models from Hugging Face and running an image registry outside of the stack. This means we are providing a complete AI system that Centria needed, with the freedom to build what they need and connect these systems to any data.

“Centria is adopting ConfidentialMind's platform to enhance project planning by securely managing sensitive data, ensuring robust compliance with data protection standards,” says Tomi Pitkäaho, RDI Manager for Centria's Production Technology Team. He adds, “The platform provides a centralized framework to systematically document and showcase Centria's expertise, enabling clear, data-driven communication of capabilities. This strategic approach strengthens Centria's networking capability and positions it as a trusted and innovative partner within its stakeholder ecosystem.”

Conclusion

The partnership with Centria demonstrates ConfidentialMind Platform ability to support diverse generative AI in education and research initiatives. Their development team was particularly excited about the platform, as evidenced by their response: "we already have some LLM based projects and are very keen to get to start using your product." The successful implementation has opened doors for expansion. For example, Centria sees potential for the platform to be used by many more people for various projects.

‍