Context And Mission
In the frame of the AIRE (“Harnessing Artificial Intelligence to TRansform Air Quality AssEssment and Management in Spain”) Spanish national project, in collaboration with CSIC Institute of Environmental Assessment and Water Research and the Univiersity of Zaragoza, the Atmospheric Composition (AC) research group within the Earth Sciences Department (BSC-ES) at the Barcelona Supercomputing Center (BSC) (www.bsc.es) is embarking on a range of cutting-edge research and development activities at the cross-section between environmental sciences (with focus on atmospheric composition and more specifically air pollution) and artificial intelligence.
In this ambitious and potentially rewarding endeavor, we are looking for a machine learning research engineer or postdoc to develop a deep learning model to emulate our BSC’s in-house MONARCH air quality model. MONARCH ingests a variety of input information including meteorology and anthropogenic pollutant emissions, to predict the spatio-temporal evolution of the atmospheric chemical composition for a range of key chemical species and aerosols. It includes a detailed representation of the main chemical and physical processes driving the fate of pollutants in the atmosphere, which comes with a high computational cost. For this specific position, we are interested in investigating the use of AI to emulate MONARCH using the Anemoi Python framework (https://github.com/ecmwf/anemoi) recently developed by the European Centre for Medium-Range Weather Forecasts (ECMWF). Anemoi is so far mainly used by European meteorological agencies to develop AI-driven weather forecasting models, whose forecasting skills are now competitive against traditional cutting-edge atmospheric models (for a much lower computational cost). In AIRE, we will extend this research to the field of air pollution. The successful candidate will work in close collaboration with other PhD students, postdocs and research engineers working on AI developments in and beyond the AIRE project. The successful candidate should also be open to directly lead or contribute AI developments in other ongoing and future research lines in the group.
See https://www.youtube.com/watch?v=EE8PijwMFXM for a general presentation of our Atmospheric Composition group. In the AC group and BSC-ES department, the successful candidate will have the chance to work in a diverse, international and highly collaborative environment. He/she will have access to the cutting-edge computational resources of Marenostrum 5, one of the largest supercomputer in Europe.
Key Duties
Acquire a deep understanding of the different components of the Anemoi framework
Learn about the different AI architectures it offers (e.g. GNNs)
Implement Anemoi on the Marenostrum 5 supercomputer
Develop AI models with Anemoi to emulate MONARCH atmospheric composition
Document and communicate the advances in a clear and comprehensive way
Contribute to other AI developments of interest for the group
Contribute to the intellectual life of the group
Requirements
Education
Postdoctoral researcher: A Ph.D. degree in environmental engineering, atmospheric chemistry, physics, climate, data science, remote sensing, computer science or similar.
Research Engineers: A Bachelor or Master degree in environmental engineering, atmospheric chemistry, physics, data science, remote sensing, computer science, telecommunications or similar.
Essential Knowledge and Professional Experience
Excellent computing skills in Python
Excellent knowledge in deep learning (e.g. CNNs, RNNs, GNNs)
Experience with Pytorch (or similar) deep learning framework
Fluency in English is essential, Spanish is optional (free lessons available after joining)
Postdoctoral researcher: demonstrated scientific expertise, including but not limited to a record of scholarly publications
Additional Knowledge and Professional Experience
Experience with UNIX/Linux and HPC environments will be valued
Experience with Git or similar software version control will be valued
Experience with coding and documentation best practices and standards
Experience with parallel programming and/or distributed training across multiple GPUs will be valued
Experience in atmospheric sciences will be valued
Competences
Very good interpersonal skills
Excellent written and verbal communication skills
Ability to organize the work, document the advancement and present results
Ability to take initiative, prioritize and work under set deadlines
Ability to work both independently and within a team