PostDoc in atmospheric physics on deep learning / foundational modelling of clouds (w/d/m)

Job description: PostDoc in atmospheric physics on deep learning / foundational modelling of clouds (w/d/m)

Background
We are seeking a scientist with a strong background in physics, meteorology or a related subject who is interested in deep learning of weather and climate processes, in particular in the atmospheric boundary layer. We also welcome applications from deep learning specialists who haveexperience working in different domains before.

The use of deep learning for weather forecasting has experienced significant growth in recent years, transforming the computational representation of atmospheric processes. Large-scale models have so far been trained on coarse-resolution model data and thereby inherited some of the weaknesses of classical models in representing important but small-scale processes in the atmospheric boundary layer. To make the next leap forward in deep learning of atmospheric processes, we intend to train a model on high-resolution model output and observations. We aim to improve deep learning models beyond the limitations of current models.

Your Tasks
You will familiarize yourselves with the deep learning techniques and models used in atmospheric and climate science. You will also fine-tune a pre-trained foundational model such as Atmorep or Aurora using diverse data from large-eddy simulations, direct numerical simulations, and observations of the real atmosphere. Furthermore, you will collaborate with atmospheric scientists in our group and machine-learning experts from the Helmholtz AI program to develop a setup that allows for training a generalizable model of atmospheric boundary-layer processes. The tasks also include presenting results at conferences and publishing them as scientific papers.

Your Profile

• PhD in physics, meteorology, or a related subject
• Strong quantitative and numerical skills (or experience with deep learning methods, particularly with transformer-based architectures from a different domain)
• Good knowledge of English as a working language (approximately equivalent to CEFR level B1)
• Prior experience in Earth system modelling and machine learning is advantageous

Further Information

The AWI is characterized by

• our scientific success - excellent research
• collaboration and cooperation - intra-institute, national and international, interdisciplinary
• opportunities to develop – on the job and towards other positions
• an international environment – everyday contact with people from all over the world
• flexible working hours and the possibility of mobile working up to 50% of regular working hours
• health promotion and company fitness with Hansefit and Wellhub
• support services and a culture of reconciling work and family
• occupational pension provision (VBL)

AWI values diversity and actively promotes gender parity, as well as an open, inclusive environment that provides equal opportunities. We are convinced that diverse teams and a variety of perspectives enrich our work and our daily collaboration. In a continuous process of learning and reflection, we aim to ensure that all our employees can be themselves and feel a sense of belonging. We welcome applications from qualified people regardless of binary and non-binary genders, race and nationality, ethnic and social background, religion, age, physical abilities, neurodivergence, sexual orientation, and other identities.

Applicants with disabilities will be given preference when equal qualifications are present.

AWI fosters work-family compatibility in various ways and has received several awards as a result of this commitment. And as a new international member of our team, you can be sure that we will help you settle in. Our Family Office and International Office will be glad to support you, even before you start at AWI.

This research center is part of the Helmholtz Association of German Research Centers. With more than 42,000 employees and an annual budget of over € 5 billion, the Helmholtz Association is Germany's largest scientific organisation.