PhD in Agentic Test-Time Adaptation for Efficient and Reliable Edge Intelligence

About the role

Fully funded four-year PhD at Aarhus University’s A3 Lab (Adaptive and Agentic AI) in the Department of Electrical and Computer Engineering. The project develops machine learning systems that adapt at test time under real-world distribution shifts, combining agentic decision-making with reliable edge deployment. Co-supervised by Dr. Behzad Bozorgtabar and Prof. Qi Zhang. Available from 1 August 2026. Deadline: 1 June 2026 at 23:59 CEST.

Responsibilities

• Design test-time adaptation methods for multimodal foundation models that maintain reliability and efficiency under distribution shifts.
• Develop agentic mechanisms that let models monitor their own confidence, detect distribution shifts, and apply lightweight self-correction.
• Evaluate methods on real edge-computing hardware with constrained memory and compute budgets.
• Collaborate within the ELLIS network and international research partners.
• Publish in venues such as NeurIPS, ICML, ICLR, and CVPR.
• Complete the Aarhus University doctoral programme coursework.

Requirements

• Master’s degree in computer science, electrical engineering, applied mathematics, or a closely related field.
• Strong background in deep learning and at least one of: distribution shift, uncertainty estimation, or continual learning.
• Proficient in Python and deep learning frameworks (PyTorch, JAX).
• Fluent English (written and oral).
• Ability to work independently and collaborate in an international team.

Nice to have

• Experience with vision-language models or large multimodal models.
• Familiarity with edge hardware (Jetson, Coral) or model compression (pruning, quantisation, distillation).
• Publications or workshop papers at NeurIPS, ICML, ICLR, CVPR, or equivalent.

How to apply

Apply via the jobs.ac.uk portal at the apply link above, or through the official Aarhus University PhD portal. Submit a motivation letter (max 2 pages), CV, academic transcripts, a writing sample (thesis draft or published paper), and contact details of two academic referees. Deadline: 1 June 2026 at 23:59 CEST.