Intelligent og effektiv dataprosesseringspipeline

Offentlig prosjektsammendrag

Our data processing software utilizes tens of thousands of sensor streams to provide customers with online decision support in the form of alerts and soft sensor estimates. While the data processing software is highly efficient, its functionality is currently limited. As a result, the algorithms devised by our data scientists are increasingly often not implementable within our framework. By providing significantly higher computational flexibiltiy while keeping efficiency high, the framework which we will develop in this project will enable modules such as computationally intensive inference algorithms and large language models to be implemented in our framework and run efficiently in real-time, forming the basis of our future AI products for real-time, data-based decision support in industrial applications. The joint development of tooling for system monitoring and management will ensure that our large-scale system remains manageable after introduction of the new functionality.