According to the Economist big data production will soon outpace available storage and the availability of computer science experts who know how to handle such data. At the same time, society is increasingly concerned about data protection. To address this gap, we need stream processing systems that continuously analyze incoming data rather than store it and allow domain experts to specify its analysis in a privacy preserving manner. The project will focus on the development of a peta-scale stream analytics system that enables domain experts to analyze high-performance data streams whilst running on a commodity cluster. The solution will support real-time advanced linear algebra operations while ensuring the privacy of the data. If successful, this project will vastly simplify the development of new, societally acceptable applications leveraging real-time data analytics.