Researcher in Machine Learning for Short-Term Forecasts

Are you passionate about research and development of AI models? Do you want to contribute to the next generation of short-term forecasts for clouds, wind, and wind power production? Are you interested in using satellite data and advanced machine learning to create more reliable forecasts that enhance societal preparedness and the stability of energy systems? Then this might be the job for you. At SMHI, you can work in a highly competent and international environment, where your expertise contributes to solutions for a more sustainable society.

About the Meteorological Research Unit

We are now seeking reinforcement for our meteorological research unit, which conducts research and development work in air quality, weather forecasting, climate analysis, and atmospheric processes. Daily work involves working with observational data, meteorological analyses, and developing models for regional weather forecasts in the Nordic countries, as well as remote sensing, climate analysis, and air quality on urban to global scales.

You can read more about the unit here: https://www.smhi.se/forskning/forskningsenheter/meteorologi (https://www.smhi.se/forskning/forskningsenheter/meteorologi)

About the Position

We are looking for a researcher in machine learning who wants to contribute to the development of AI-based short-term forecasts for meteorology. The work involves developing and training dedicated ML models that are integrated with the open atmospheric model WeatherGenerator and supplemented with data from satellites, radar, aircraft, and wind turbines. WeatherGenerator is a generative foundation model that provides a compact description of the atmosphere. SMHI is developing it together with European partners within an EU project, where the model is intended to serve as a digital twin in the EU Commission's initiative Destination Earth. The position is linked to a project funded by the Swedish National Space Agency within the Climate and Environment 2025 call, where SMHI and Svenska kraftnät are developing a new system for fast and high-resolution probabilistic forecasts of clouds, wind, and wind power production.

You will:

• Develop and train dedicated ML models for short-term forecasts of cloud and wind characteristics
• Integrate satellite data (AVHRR, VIIRS, AWS, etc.), radar, aircraft data, and observations from wind turbines
• Work in a shared codebase compatible with the WeatherGenerator project (PyTorch, Python, GitHub)
• Participate in evaluating the forecasting models against existing systems
• Contribute to scientific communication, publications, and knowledge dissemination

The work is carried out in close collaboration with SMHI's researchers in remote sensing, radar, nowcasting, and ML, as well as with PhD students at Svenska Kraftnät and the project's international partners within the EU project WeatherGenerator (https://player.vimeo.com/video/1050751131 (https://player.vimeo.com/video/1050751131)).

About You:

We are looking for someone who is an engineer/physicist/mathematician/statistician/data scientist. A PhD is required.

Requirements:

• Experience in ML development in Python with PyTorch
• Good experience working in a Linux/Unix environment, including HPC clusters
• Ability to write efficient, testable, and maintainable code in collaboration with others via GitHub
• The work should be performed/checked in a team. Good collaboration skills combined with determination and initiative are required.
• Good ability to communicate results in internal and external contexts.

Desirable:

• Experience in documenting and reporting work both technically and scientifically
• Experience working with shared codebases in international projects and interdisciplinary collaboration
• Particular consideration will be given to education/knowledge/interest/experience in the following areas as desirable:
• Applied machine learning in geoscientific domains using, e.g., representational learning, deep learning, probabilistic modeling
• Mathematical statistics: e.g., Bayesian inference
• Satellite data and remote sensing
• Handling large datasets in HPC environments

As a person, you are challenged by problem-solving and have good analytical skills. Curiosity and a desire to continuously develop your skills in a rapidly changing research field are prerequisites. You are proactive and take responsibility for your work, which you carry out in a structured manner to deliver requested results on time. You are goal-oriented and set and adhere to deadlines.

You have good knowledge of English in both spoken and written form. If you lack knowledge of Swedish, it is important that you can use Swedish as a working language within a year.

Welcome with your application

Employment type: The position is a two-year fixed-term employment, with the possibility of a one-year extension and starting as soon as possible.

Location: Norrköping

Application deadline: 2026-01-19

Read more about SMHI as an employer and about our offerings here.

SMHI is a Swedish expert authority with a global perspective and a vital task in forecasting changes in weather, water, and climate. Based on scientific principles and through knowledge, research, and services, we contribute to increasing the sustainability of society as a whole. Every day, around the clock, year-round.

SMHI is a preparedness authority, and as an employee, you may be assigned to the authority in the event of war.

Keep in mind that the documents and information you send to SMHI through your application become public records. This means that all material in the application, including attachments, may need to be disclosed to anyone who requests it unless the information is covered by confidentiality under the Public Access to Information and Secrecy Act. Please consider writing primarily what you deem relevant in relation to the requirements of the position. Be mindful of your privacy and avoid providing information that contains sensitive personal data, information about your or a close relative's health, political opinions, or religious beliefs.