Postdoc: Physics‑Driven ML for TEM Subsurface Imaging

Signal Processing and Machine learning - Edison; Denmark; Hybrid; DKK 480,000 - 560,000; Full time

Benefits offered by this job

Relocation support

Flexible work culture

International collaboration

Mentoring

Job summary

Signal Processing and Machine learning - Edison at Aarhus University invites applications for a two-year postdoctoral position focused on physics-driven machine learning for TEM data used in subsurface imaging. The role centers on integrating electromagnetic physics into supervised and self-supervised learning within the anEMone framework.

Responsibilities include developing and validating ML models, creating physics-based loss functions, incorporating system geometry and measurement

Qualifications

• PhD in a related field (see above) with strong publication record.
• Proven experience in deep learning model development (CNNs, UNets, transformers).
• Strong programming skills in Python and PyTorch.

Responsibilities

• Develop physics-driven ML models for TEM inversion.
• Design physics-informed loss functions.

Skills

Deep learning

Scientific ML

Inverse problems

Python

PyTorch

CNNs

UNets

Transformers

English communication

Education

PhD in Electrical Engineering

PhD in Computer Science

PhD in Physics

Tools

PyTorch

NumPy

Git

Linux

Job description

Signal Processing and Machine learning - Edison at Aarhus University invites applications for a two-year postdoctoral position focused on physics-driven machine learning for TEM data used in subsurface imaging. The role centers on integrating electromagnetic physics into supervised and self-supervised learning within the anEMone framework.

Responsibilities include developing and validating ML models, creating physics-based loss functions, incorporating system geometry and measurement