Thomas Moscibroda is a Distinguished Engineer at Microsoft. He leads the Azure Core Platform Capacity & Efficiency Team and is responsible for resource allocation and optimization, workload placement, capacity management, and efficiency in the Azure Core platform. In this role, Thomas drives initiatives to improve Azure’s Core margin efficiency and capacity experience. Many of Thomas inventions and technologies have shaped key foundational services at the heart of the Azure cloud, including for example Azure’s allocator service, its buffer management and admission control systems, or its capacity management system (CMAS).

Thomas’ academic research interests are in cloud computing, distributed systems, networking, and most recently blockchain technology. He has a particular focus on algorithmic and mathematical approach to practical system problems. His research is documented in more than 100 internationally peer-reviewed research papers and has led to more than 50 patents. His work on the foundations of distributed systems and computing, and on white-space networking have received Best Paper Awards from several prestigious academic publication venues, including ACM SIGCOMM 2009, USENIX NSDI 2009, ACM ASPLOS 2010, EuroSys 2012, ACM/IEEE IPSN 2007 and 2017, ACM PODC 2004 and 2012, DISC 2015, as well as MLSys 2023. His articles on DRAM scheduling and on-chip networking in multi-core systems were selected as IEEE Micro Top-Pick Computer Architecture papers in 2008 and 2010, respectively; and his article on ML-centric cloud platforms and its use in capacity-efficient resource management in Azure was featured in the Communications of the ACM. Thomas is also the recipient of the MICS Research on Communications Award by the National Research Foundation of Switzerland (NCCR) for his contributions to the area of Mobile Communications & Information Systems. Thomas has served on numerous technical program committees, including for SOSP (2017), OSDI (2016, 2023), NSDI (2016), PODC (2009, 2010, 2015), MOBICOM (6 times), EuroSys (2015), ICALP (2013, 2016), and IPSN (2012, 2013).