netidee Scholarships: TÜV Austria Wissenschaftspreis
Daniel May was honored by TÜV Austria for his thesis on energy-efficient edge AI at TU Wien.
The master's thesis on DynaSplit, which deals with energy-efficient AI inference along the edge-cloud continuum, was awarded the TÜV AUSTRIA Wissenschaftspreis 2025 in the category of diploma/master's theses.
The master's thesis submitted to the Vienna University of Technology received the TÜV AUSTRIA Wissenschaftspreis 2025 in the category “Universities/Universities of Applied Sciences – Diploma/Master's Theses.” The winners were announced on November 24, 2025, at an event at the Vienna University of Technology.
The thesis addresses the question of how split computing can be used to increase the efficiency of AI inference in edge cloud environments. The goal is to reduce energy consumption while meeting specified latency requirements. To this end, the thesis examines how neural networks can be sensibly divided between edge devices and cloud resources and how targeted adjustment of hardware parameters contributes to optimization.
The results of the evaluation show that the optimal execution strategy depends heavily on the respective application context. In particular, the model architecture, the required latency, and the available resources have a significant influence on whether execution at the edge, in the cloud, or a split approach delivers the best results. These findings highlight the importance of systematic, query-dependent configuration for practical use.
December 2025
More information