Jan: "Most systems in Big Data analytics are operating on high-level abstractions, far aware from the hardware, and therefore not very efficient. They are designed for a heterogeneous environment, when in reality you only have one or a few types of machines in your data center and in most cases you know the infrastructure. When you are aware of the topology, the network, and the machine design, you can built a platform that is faster and more efficient in dealing with unstructured, ad hoc or high volumes of data. I liked my job at IBM Research in Austin, Texas, but I wanted to go back to the academic world full-time, after spending six years in the industry. I studied at ETH Zurich and I knew TU Delft through the Idea League. It is an added bonus that I am about two hours away from my home town in Germany now."
