PhD Stipend in Energy-Efficient Agentic AI and Sustainable Computing Aalborg University in Denmark

Aalborg University invites applications for a three-year PhD project on the energy efficiency and responsible use of agentic artificial intelligence. The position is hosted by the Department of Computer Science in the Distributed Sustainable Software and Systems (DSSS) research group, part of the Distributed, Embedded and Intelligent Systems (DEIS) section.

The project investigates when the use of AI is worth its energy cost. Agentic AI systems may plan, call tools, retry operations, and check their own results, potentially consuming far more energy than ordinary conversational use. The PhD researcher will build an evidence base for measuring energy consumption per successfully completed task and determining when more energy-efficient alternatives can perform the same work.

A central research comparison is between repeatedly invoking an AI agent at runtime and generating conventional deterministic software once. The project will study the break-even point between these approaches and how it varies with the model, framework, task stability, and development practices. Potential research directions include measuring energy use in large language model and agentic-coding workloads, analysing energy spent on planning and failed steps, evaluating prompt caching and model routing, estimating energy consumption for API-based models, and assessing how provider and model changes affect system stability and portability.

The work is experimental. The researcher will run agent frameworks on curated task sets and measure energy using tools such as Intel RAPL, performance counters, and rack-level power meters. For frontier models, the project will develop and validate telemetry- and FLOP-based estimates. The research environment provides energy-measurement testbeds, on-premise GPU and CPU infrastructure, and access to Aalborg University's CLAAUDIA cloud and HPC services.

The project is part of Energy, Sovereignty and the Responsible Use of Agentic AI, funded by the Independent Research Fund Denmark. It is a collaboration between Aalborg University's Department of Computer Science and Business School. The PhD researcher will work closely with social-science researchers, participate in shared seminars, undertake a short secondment in the organisational work package, and contribute to pilots with Danish public and private organisations.

The project follows open-science practices, including preregistered protocols, reporting of null and negative results, and release of measurement harnesses and benchmarks as open source. Research findings are intended for venues in software engineering and sustainable computing, while the researcher will also join the department's problem-based and internationally oriented academic environment.