Data Scientist for Weather Forecasting using Machine Learning

This project arises from a collaboration between the Swiss Data Science Center (SDSC), MeteoSwiss, and EUMETNET. 

The SDSC has been a National Research Infrastructure since 2025, evolving from a strategic focus area of the ETH domain, with EPFL and ETH Zurich as founding partners. Its mandate is to support academic groups and research, hospitals, industry, and the public sector at large, including cantonal and federal administrations. The center accompanies and supports their entire data science journey, from the collection and management of data to machine learning, AI, and industrialization. As part of the mission of the SDSC, we are also tackling problems arising in several domains: Climate, Weather, and Environmental Sciences is a vertical within SDSC of which we strongly contribute. The Center comprises a multi-disciplinary team of data and computer scientists and experts in several domains, with offices in Zürich, Lausanne, and Villigen.

MeteoSwiss, the Federal office for Meteorology and Climatology, leads and coordinates a larger effort on ML-based weather forecasting together with its European partners, including other national weather services and ECMWF. In its role, MeteoSwiss will provide domain science support, access to data and know-how, and will support the onboarding onto ongoing efforts and existing codebases. MeteoSwiss is also responsible for prototyping and operationalization of the developments done in the context of this project. As such, it defines requirements and provides feedback on the outcomes.

The advertised position is funded by EUMETNET (the European Meteorological Network) programme on Artificial Intelligence and Machine Learning for Weather, Climate and Environmental Applications (E-AI). EUMETNET is a network of the 33 European National Meteorological Service, which exists to provide a framework to organise co-operative programmes between the members in fields of meteorology, data processing and forecasting products.

We are looking to hire a data scientist with expertise in machine and deep learning for the development of a seamless neural weather forecasting system. This project is framed as a collaboration involving the Swiss Data Science Center, MeteoSwiss, and EUMETNET, and it is funded for 2 years.

The goal is to develop and implement a system able to perform forecasting at various temporal scales up to 10 days, using recent developments in deep learning and machine learning. We expect the candidate not only to contribute to methodological advances, but also to develop and contribute to existing codebases and open-source initiatives in the domain, with good quality, reproducible, and robust code in Python. The candidate will make use of and contribute to Anemoi, a toolbox for the use and implementation of AI tools for weather and climate applications, developed by the European Center for Medium-range Weather Forecast (ECMWF).

Another important project goal is to assess the integration of such systems into operational forecasting services, complementing numerical and statistical models currently used at MeteoSwiss but ideally at other European Weather Services. 

The candidate is also expected to participate in frequent meetings and updates with relevant stakeholders, such as MeteoSwiss and other EUMETNET members and advisory committees. 

The position is located at the Swiss Data Science Center, allowing the candidate to interact with a wide variety of experts in the machine learning and data science domains. Our diverse and interdisciplinary environment offers the candidate the opportunity to engage in all types of research discussions, and advance in the methods and the domains fields alike.

Main duties and responsibilities include:

• You will focus on collecting data and streamlining data products, in collaboration with MeteoSwiss and relevant European data providers, and coding a dedicated anemoi-dataset parsing routine for efficient training and inference purposes
• You will review and implement baselines and state-of-the-art methods relevant to the task, and critically assess them
• You will develop, in collaboration with MeteoSwiss scientists and engineers, the weather forecasting models, the core of this project 
• You will use and implement assessment and verification tools, interpret them, and communicate relevant results to stakeholders
• You will complement and contribute to the ANEMOI library, where relevant (baselines, assessment, own methods, own functionalities)
• You will engage with the relevant community proactively

For this position, we are looking for a young scientist with a completed MSc thesis in a relevant domain. 

• You have a MSc in computer science, machine learning, data science, or related field, and have worked with geospatial data, weather data, climate data
• You have experience with or knowledge of both traditional and modern machine learning: from training a random forest to training a transformer graph neural network, with experience on the latter being considered an advantage.
• You are strongly motivated to work in a dynamic research environment and to invest yourself in an exciting research project.
• You are familiar with recent literature and follow recent advances in the field
• You are comfortable in coding with Python, and have experience working on large collaborative and open source codebases. You know git and best practices for collaborating on multidisciplinary and scientific code. Preferentially, you know community standards and practices for doing so. 

• We consider familiarity with Pytorch, Pytorch-Lightning, hydra, zarr and xarray, and earthkit a plus. You can understand external complex libraries such as Anemoi, and know how to collaborate on such codebases.
• You have experience in scientific research (e.g. acquired during the MSc thesis), specifically, in presenting results in a concise and goal driven manner and in presenting to scientific and non-scientific partners.
• You have experience working in diverse interdisciplinary teams.
• You are eager to work in an agile setup

• A stimulating, collaborative, cross-disciplinary environment in a world-class research institution, where you will be part of a team of 40 data scientists from more than 15 different countries. We all work towards applying and developing novel ML methods to solve real-world problems.
• Close collaboration with a dedicated ML team at MeteoSwiss, including regular visits to their main office at the Zurich Airport
• We value work-life balance and support home-office work days
• You will have a budget for travel and attending events and conferences
• You will be able to publish your research in collaboration with experts from different fields in top-ranked journals and conferences
• You will have the opportunity to help supervise student projects
• We encourage experimentation and creativity by actively promoting learning of new technologies and approaches on the job
• Your opinion will always matter 
• Salary is in accordance with ETH Zurich regulations