Philipp Wiesner

I’m a researcher in the Cambridge Machine Learning Systems Lab at the University of Cambridge. Previously, I was a postdoctoral researcher in the Distributed and Operating Systems (DOS) group at TU Berlin, where I completed my PhD with distinction in June 2025. Before and alongside academia, I spent several years at startups, most recently working on distributed AI systems.

Vision & Interests

Datacenters are becoming gigawatt-scale consumers of electricity and embodied resources, but their environmental impact remains difficult to measure and control. My research develops methods to quantify and reduce this impact in large-scale computing and AI systems. I also study how datacenters can use their flexibility to stabilize the electrical grid and accelerate decarbonization.

I believe sustainable computing will require more than efficiency, and that computer systems research has an important role in informing policy and governance.

I’m interested in research at the intersection of computing and energy systems, such as carbon-aware computing [Middleware'21, e-Energy'24, e-Energy'25], modeling and analysis [HotCarbon'24, SSC'25], and efficient ML systems [NeurIPS'24, HyperAgents'25].

• September 2026: I joined the Cambridge Machine Learning Systems Lab as a postdoctoral researcher, where I will lead a new research direction on sustainable, large-scale AI systems.
• August 2026: Our benchmarking paper on Evaluating MFU as a Proxy for GPU Power was accepted at MASCOTS'26.
• April 2026: I gave my first keynote talk on “Co-Simulating Computing and Energy Systems” at the TDIS workshop co-located with ACM EuroSys'26 in Edinburgh.
• March 2026: We’ll present our latest research on distributed LLM pretraining during renewable curtailment windows at the Flower AI Summit in London.
• February 2026: Over the next four months, I will be visiting Loïc Lannelongue at the Cambridge Sustainable Computing Lab to work on energy and carbon accounting in computing.