Das Deutsche Klimarechenzentrum (DKRZ) ist die zentrale Simulations- und Datenverarbeitungseinrichtung für die deutsche Klima- und Erdsystemmodellierung und gehört zu den führenden Einrichtungen in diesem Bereich. Das DKRZ betreibt nicht nur Supercomputer der höchsten Leistungsklasse und eines der größten Daten- und Archivsysteme weltweit, sondern beteiligt sich auch an vielen nationalen und internationalen Projekten zur Verbesserung der Software und Infrastruktur für die Klimamodellierung.

Momentan bieten wir eine offene Stelle als

Post-Doctoral Computational Research Scientist (all genders)

On the exciting fast evolving interface of climate research and artificial intelligence / machine learning (AI/ML), we are seeking a Computational Research Scientist with a passion for working at the cutting edge of AI/ML, to bring new capabilities to climate science. You will join a high impact team of scientists in the Climate Informatics and Technology group at DKRZ, working on deeply technical problems in direct collaboration with leading academics from various partners across Europe.

The position (100% - until September 2025 - subject to financing through BMBF) is within the German Ministry for Education and Research (BMBF) project ExaOcean “Performance improvement of the ICON-O ocean model on heterogeneous exascale supercomputers using machine learning methods.”

The goal of ExaOcean is to significantly improve the execution times and parallel scalability of the ICON-O ocean model by optimizing numerical algorithms using machine learning techniques. The DKRZ is concerned with the development of a cross-project interface technology for the effective calculation of a "super-resolution" (SR) approach on high-performance computer architectures. For this purpose, an image processing routine of the so-called Image Inpainting using machine learning will be scientifically adapted for the requirements of the project in such a way that it can read in ICON-O ocean model data, process and evaluate them for the determination of eddies and accordingly provide a high-resolution database.

Job description:

  • Machine Learning: installation on HPC (CPU/GPU) infrastructure, advance established technology regarding workflow, I/O, training and validation of the neural networks, optimized network set up incl. convolution of grids and its flexibility
    • https://github.com/FREVA-CLINT/climatereconstructionAI)
  • Climate Science: advance established technology for super-resolution or downscaling, analysis of ocean states, eddy detection, scientifically valid input and output files, preparation of data sets from and for partners
    • Kadow et al. 2020 - https://rdcu.be/b4ysS
  • Software Development: Provision of the super-resolution technology as library that can be called by an interface offline and potentially online during an ICON-O model run
    • https://mpimet.mpg.de/en/science/modeling-with-icon/icon-configurations/icon-o
  • Project related communication, cooperation and scientific interaction with ExaOcean and related partners for overall strategies

Requirements:

    • A PhD is a plus. University degree (MSc) in Computer Sciences and/or Natural Sciences (climate, ocean, atmosphere), or similar fields is mandatory.
    • Programming skills, as well as profound knowledge and experience in the field of machine learning
      • Python3 and PyTorch, neural network improvement, hyperparameter tuning

    Desirable:

    • Experience in the field of climate model data, ocean modeling
      • data standards, evaluation of climate models, statistics, scientific programming
    • Knowledge in software development in cross-language development
      • High Performance Computers (Unix/Linux), Bash, Job-Scheduling, GitHub/GitLab

    This is an exciting opportunity in a flexible academic environment where you have the chance to develop initiative and shape services of the next generation in an area of high social relevance. We expect capacity for teamwork in an interdisciplinary environment, the willingness to travel nationally and internationally, and fluent English language skills. German is a plus. At DKRZ, we recognize that training is vital for developing staff so you will have access to courses to increase the depth and breadth of skills you have. DKRZ is a family friendly environment, allows flexible working hours and has Home Office options.

    Payment will be in accordance with the German public service positions up to TVöD/Bund E13 according to the applicants’ qualifications, including extensive social security plans. The conditions of employment follow those of the German civil service.

    • https://oeffentlicher-dienst.info/tvoed/bund

    How to submit your application

    Please submit a letter of interest and detailed curriculum vitae in a single pdf file. Please send your application by email with the reference DKRZ_15_2022 in the subject to bewerbungen@dkrz.de.

    Deadline for applications

    This vacancy has been opened 29-August-2022. Review of applications will commence after 30-September-2022. For further information, please contact Christopher Kadow (a2Fkb3dAZGtyei5kZQ==).

     

    Deutsches Klimarechenzentrum GmbH
    Personalverwaltung
    Bundesstraße
    20146 Hamburg