Förderjahr 2024 / Stipendium Call #19 / ProjektID: 7170 / Projekt: Neural network splitting for energy-efficient Edge-AI
Award update: my master’s thesis on DynaSplit, energy-efficient AI inference across the edge-cloud continuum, won the TÜV AUSTRIA Science Award 2025 in the Diploma/Master category.
A milestone for my thesis work
I’m happy to share that my master’s thesis at TU Wien received the TÜV AUSTRIA Science Award 2025 in the category “Universities / Universities of Applied Sciences – Diploma/Master Theses.” The winners were announced on 24 November 2025 at TU Wien.

(c) TÜV AUSTRIA Group/APA-Fotoservice/Richard Tanzer
Official announcement: https://www.tuv.at/tuev-austria-wissenschaftspreis-2025-die-sieger-stehen-fest/
What the thesis is about
My thesis explores how split computing can make AI inference more efficient in edge-cloud environments. The central goal is to reduce energy consumption while still meeting latency requirements, by deciding which parts of a neural network should run on the edge and which parts should run in the cloud, and by tuning hardware parameters accordingly.
What the results indicate
The evaluation indicates that the best execution strategy depends strongly on the context: the model architecture, the latency target, and the available resources all influence whether edge execution, cloud execution, or a split setup performs best. This underlines why systematic, request-aware configuration is important for practical deployments.
Thanks
A big thank you to everyone who supported this work, especially my supervisors, colleagues, friends, and family. I’m also grateful to the organizers and jury of the TÜV AUSTRIA Science Award for this recognition.