RSE HPC Programming (all genders)

Overview

The German Climate Computing Centre (DKRZ) is the central simulation and data processing facility for the German climate and Earth system modelling community and a leading institution in this field. DKRZ operates supercomputers, a large data and archive system, and participates in national and international projects to improve software and infrastructure for climate modelling. We are offering a position as

Research Software Engineer (all genders) for Earth System Modelling

We are looking for new colleagues who are passionate about high-performance computing and developing research software. Would you enjoy working in a team that improves modern climate models? Do you have experience adapting software for use on existing and future supercomputers? Would you like to apply your knowledge of AI to science and support others in this field? If so, DKRZ could be the right place for you.

Tasks

As part of the Earth System Modelling support team, we are committed to pushing the boundaries of Earth System Modelling (ESM) and climate research. Joining our team offers the opportunity to work closely with other highly skilled Research Software Engineers (RSEs) in an environment that encourages flexibility and innovation. The RSEs are strategically located between German and European HPC centres and colleagues in scientific research. The working day includes:

• Collaboration with Earth system scientists and climate researchers, addressing technical challenges in HPC, AI and software development
• Provision of expertise as a service to scientists over a period of up to 6 months, adapting to evolving needs and fostering dynamic solutions
• Working closely with other RSEs and scientists, ensuring efficient project implementations within tight timeframes, focusing on ESM development for current and future HPC systems

To be successful in this role, you should have a strong interest in interdisciplinary Earth system modelling and be keen to apply your HPC or AI programming expertise to take on new challenges.

Qualifications / Experiences

As the requirements of our project partners and customers are varied, we do not expect applicants to meet all of the following criteria. A clear profile and focus on a specific topic can also be very helpful.

• Academic degree in computational science, applied mathematics or equivalent training and knowledge
• Strong experience in parallel programming for HPC systems (MPI, OpenMP)
• Experience in GPU programming using standards like CUDA, HIP or OpenACC
• Experience in using machine learning methods in a scientific HPC environment
• Confident handling of tools such as Git, Spack, shell scripting
• Very good communication skills and knowledge of English

Employment conditions

We offer work in interdisciplinary teams and excellent connections to national and international research networks. In addition to specific implementation and support requirements from our users and project partners, your work will be characterised by the freedom to work creatively at the interface between applied computer science and the development of scientific applications. Participation in international conferences, project meetings and hackathons enables further training and presentation of your own results.

We are based in Hamburg, but offer flexible working hours and location (up to 100% work from home within Germany). Remuneration will be based on the candidate's qualifications according to TVöD-Bund. You can start in November 2025, but the exact starting date can be agreed individually. The position is initially limited to two years, with the option of a further two years following a successful review by the funding body.

Application process

Interested applicants are requested to compile a single PDF document containing a cover letter, a detailed CV and any additional documents (e.g., professional references). Applications must be submitted via the portal: dkrz.softgarden.io/en/vacancies

Application deadline: Open until filled. Priority will be given to applications received by 30 September 2025.

Questions? Hendryk Bockelmann ( ) will be happy to answer them in advance.