Presentation of the habilitation project
Wednesday, November 04, 2026 | 01:30 p.m. | Room: N.0.43 | University of Klagenfurt
Dr. Reza Farahani
Distributed System Groups (DSG) | TU Wien | Austria
Abstract: The growing adoption of AI-based services and applications is reshaping service management across the edge-cloud continuum. Building systems for AI requires coordinating heterogeneous computational resources to meet requirements on latency, accuracy, and cost. Meanwhile, AI for systems introduces predictive models, learning algorithms, and agentic mechanisms to support resource and service management. Both directions require service execution to adapt to changing demand and infrastructure conditions, extending the management problem beyond capacity provisioning. Execution placement, model configuration, and application-specific processing also influence resource demand, delivered quality, monetary cost, and energy consumption, motivating a multidimensional view of elasticity. This talk first examines these interdependencies through resource, cost, and quality elasticity. It then presents research contributions addressing complementary aspects of elastic service management, from deadline- and energy-aware service orchestration to quality-aware AI inference and intent-driven processing. It further explains how empirical profiling and predictive models support optimization and reinforcement learning for placement, resource allocation, and service configuration. Through experiments involving language and vision-language model services, video processing/streaming, and UAV inspection on heterogeneous edge-cloud infrastructures, the talk illustrates the trade-offs, effectiveness, and limitations of these approaches. It concludes by discussing open challenges in coordinating these adaptation mechanisms across infrastructure and application layers.
Bio: Reza Farahani is a Univ.Ass. Postdoctoral Researcher and Lecturer at the Distributed Systems Group (DSG), TU Wien, Austria, and an External Lecturer at the University of Klagenfurt. Previously, he was a Postdoctoral Researcher and Lecturer at the Department of Information Technology (ITEC), University of Klagenfurt (2023-2026), where he also earned his Ph.D. in Computer Science. During his doctoral research, he contributed to the Christian Doppler Laboratory ATHENA and was a Visiting Scholar at the University of Surrey’s 5G&6G Innovation Center. Dr. Farahani has co-authored 60+ peer-reviewed publications and is a co-recipient of the Distinguished Paper Award at IEEE HiPC 2024. He coordinates the FFG-funded EdgeAI-Drone project and led a Green AI research project under the EU ENFIELD Exchange Scheme in collaboration with SINTEF, Norway. Previously, he served as a tool leader and WP5 co-leader in the Horizon Europe Graph-Massivizer project, contributing to sustainable service orchestration and the development of a federated HPC-Edge-Cloud experimental testbed. He leads the Advanced Internet Computing course at TU Wien, with 300+ master students, and the Serverless Distributed Systems master course at the University of Klagenfurt. He serves as an Associate Editor of IEEE Network Magazine and a Guest Editor for Future Generation Computer Systems. His conference leadership roles include Area Chair for ACM Multimedia 20254-2025, TPC Co-Chair of iEDGE 2026, Track Chair for ACM Web 2027, Publicity Co-Chair for IEEE CCGrid 2027, and IEEE ICFEC 2027. He has also delivered tutorials at IEEE/ACM UCC 2025, IEEE ICDCS 2025, ACM/SPEC ICPE 2025, and EUVIP 2024. His research focuses on intelligent and sustainable service management across the edge-cloud continuum, spanning distributed and networked systems, serverless computing, edge and agentic AI, LLM/VLM inference orchestration, distributed multimedia systems, and multi-objective optimization.
