Data scientist AI driven weather model

Koninklijk Nederlands Meteorologisch Instituut; De Bilt; On-site; EUR 65,000 - 90,000; Full time

Job summary

KNMI invites a data scientist to join the Data Science Cluster in the RDWK department, contributing to the DE_376 Destination Earth project. You will develop state-of-the-art ML models for data-driven weather forecasting, focusing on post-processing NWP outputs and stretched-grid models using ERA5 and high-resolution reanalysis data.

Collaborate with European institutes to benchmark against Harmonie-Arome and advance probabilistic forecasts of extreme weather, exploring diffusion models and

Qualifications

• MSc in data science or related field, strong focus on machine learning.
• Hands-on experience developing deep learning models using PyTorch.
• Experience in weather forecasting models is a bonus.
• Experience with data-driven weather models (e.g. ERA5) is a bonus.

Responsibilities

• Develop state‑of‑the‑art ML models for data‑driven weather forecasting.
• Contribute to stretched‑grid weather model using graph neural networks and graph transformers.
• Train and optimize models with large weather datasets like ERA5 and high‑resolution reanalysis data.
• Evaluate forecasts against Harmonie‑Arome and benchmark performance.
• Collaborate with European institutes on DE_376 for probabilistic extreme weather forecasting.
• Explore diffusion models to improve next‑generation weather prediction.

Skills

PyTorch

Education

MSc in data science or related field

Job description

Are you passionate about advancing weather modeling using cutting‑edge technology? Join the Royal Netherlands Meteorological Institute (KNMI) in the Destination Earth (DE) AI project (DE_376), where we push the future of high-resolution weather forecasts through deep learning. We are looking for a data scientist to help develop a model that matches the accuracy of our Harmonie‑Arome model, especially for extreme weather. Collaborate with top meteorological institutes across Europe and shape the next generation of weather modeling.

How you contribute

As part of the Data Science cluster in the R&D Weather and Climate Models department, we are driving the development, maintenance, and application of several cutting‑edge machine learning models. Our focus is on advancing ML methods for post‑processing of numerical weather prediction (NWP) model output, as well as further developing data‑driven weather models using machine learning – a task where you will play a key role. You’ll contribute to the further development of a stretched‑grid weather model, leveraging a 40‑year ERA5 re‑analysis archive and multi‑year high‑resolution (km‑scale) re‑analysis data from several NWP models, such as our Harmonie‑Arome model. This model is a graph neural network with a graph transformer.

Additionally, you will compare the performance of the forecasts from the stretched‑grid weather model with those from Harmonie‑Arome.

This exciting work is part of the DE_376 project. You’ll be collaborating with leading European meteorological institutes to develop high‑resolution data‑driven weather models for Europe with a focus on improving probabilistic forecasts of extreme weather by incorporating diffusion techniques a.o.

Your activities

• You develop and improve state‑of‑the‑art machine learning models for data‑driven weather forecasting.
• You contribute to the development of a stretched‑grid weather model based on graph neural networks and graph transformers.
• You train and optimize models using large‑scale weather datasets, including ERA5 and high‑resolution NWP reanalysis data.
• You evaluate and benchmark machine learning weather forecasts against the Harmonie‑Arome numerical weather prediction model.
• You collaborate with leading European meteorological institutes on the DE_376 project to advance probabilistic forecasting of extreme weather.
• You explore and apply innovative AI techniques, such as diffusion models, to improve next‑generation weather prediction.

Your team

You will join the Data Science Cluster within the R&D Weather and Climate Models Department at KNMI. The cluster brings together data scientists and researchers who develop innovative machine learning solutions for weather and climate applications.

In the Destination Earth (DE_376) project, you will directly collaborate with 2 other colleagues at KNMI, while also working with other leading European meteorological institutes. The team combines scientific excellence with an open and collaborative culture, where knowledge sharing and innovation are central.

Would you like to know more?

We understand that you might want to learn more about this position. Feel free to contact Maurice Schmeits, coordinator of the Data Science cluster, at [phone] or Ben Wichers Schreur, team leader, at [phone]. They will be happy to assist you!

KNMI's unique task is the gathering of information about the atmosphere and the subsurface and the translation of that information to risks for society.

Royal Netherlands Meteorological Institute (KNMI)

The weather is temperamental, the ground moves and the climate changes. For our safety and prosperity, we need to know what risks and opportunities this brings. And: how we can best prepare ourselves. The Royal Netherlands Meteorological Institute (KNMI) is the national knowledge and data centre for weather, climate and seismology. Reliable, independent and focused on what the Netherlands needs. For a safe Netherlands that is prepared for the impact of weather, climate and earthquakes.

We use our core values – Development, Cooperation and Relevance – to achieve our ambition, both within and outside KNMI, nationally and internationally. We develop our knowledge and expertise and work together to create a single KNMI that makes a difference to society!

Organizationally, KNMI falls under the Ministry of Infrastructure and Water Management. The Ministry of Infrastructure and Water Management (IenW) is committed to a safe, accessible and livable Netherlands. That is why the Ministry is working on powerful connections by road, rail, water and air. And IenW protects against flooding, ensures the quality of air, water and soil and the realization of a circular economy.

Talent as the basis, diversity as the strength

The KNMI is an inclusive organization. An organization that provides space for everyone and uses the strength of its diverse workforce to achieve better results together for the Netherlands. Inclusive means that everyone feels involved and valued; not in spite of their differences, but thanks to them.

KNMI: R&D Weather and Climate models

The department Research and Development of Weather and Climate models (RDWK) investigates and develops research tools for weather and air quality prediction applications and climate research. We work on detailed physical processes, data assimilation, long term climate projections and practical applications including storm surge forecasts and statistics of extremes. RDWK participates in international projects directed towards a variety of weather and climate related research and development areas. The department has a strong international network.

The department performs its activities in collaboration with partners like ACCORD, EC‑Earth, the European Centre for Medium‑Range Weather Forecasts and universities. We work for the national road and water authority Rijkswaterstaat, the Ministry for Infrastructure and Water Management, the Dutch research funding organisation (NWO), the European Copernicus programme and other European research programmes; and on servicing the operational weather forecast centre, climate scenarios and strategic research. The department collaborates with the other KNMI departments.

Qualifications

• You hold an MSc in data science or a related field, with a strong focus on machine learning.
• You have hands‑on experience developing deep learning models using PyTorch.
• Experience in weather forecasting (models) is a bonus.
• Experience with a data‑driven weather model (like AIFS of ECMWF) and/or Anemoi is a bonus.

We are looking for someone who demonstrates the following competencies

• Analytical skills: you are good in interpreting and analyzing scientific results.
• Collaboration: you enjoy working in multidisciplinary and international teams.
• Results orientation: you focus on delivering high‑quality outcomes and achieving project goals.
• Innovation: you are eager to explore and apply new machine learning techniques.
• Communication skills: you communicate complex ideas clearly, both verbally and in writing.

• We have opportunities for two full‑time positions (2 FTE) within this project.
• These are temporary positions. Depending on the position offered, the employment contract will be for 15 or 18 months.
• A valid EU work permit is required.
• Once we have found the right candidates, we close the vacancy.
• Starting date: January 1 2027 – we’re eager to have you join us.
• Obtaining references can be part of the application process. An (online) pre‑selection test, an assessment, making a case and (online) screening via databases, among others, can be part of the application process.
• Acquisition in response to this vacancy is not appreciated.