Research Engineer / Postdoc to develop physically consistent super-resolution downscaling deep [...]

Barcelona Supercomputing Center; Barcelona; Presencial; EUR 36.000 - 60.000; Jornada completa

Descripción de la vacante

The Barcelona Supercomputing Center is seeking a dedicated Research Engineer or Postdoctoral Researcher for the Atmospheric Composition group. This role focuses on developing deep learning models for atmospheric composition predictions, requiring strong qualifications in computer science and applied mathematics, along with expertise in deep learning and high-performance computing.

Servicios

Good working environment

Flexible working hours

Extensive training plan

Private health insurance

Support for relocation procedures

Restaurant tickets

Formación

• Strong background in deep learning and geographical sciences.
• Experience with ML frameworks and high-performance computing.
• Excellent programming skills in Python.

Responsabilidades

• Develop deep learning architectures for atmospheric models.
• Train and validate models using historical datasets.
• Collaborate with atmospheric scientists and contribute to publications.

Conocimientos

Deep Learning

Image Super-resolution

Geospatial Applications

Python

HPC

Interpersonal Skills

Fluency in English

Educación

PhD or MSc in related discipline

Herramientas

PyTorch

TensorFlow

Git

Descripción del empleo

We are looking for software atmospheric modeler to join the Atmospheric Composition group within the Earth Sciences department at the BSC-CNS. Composed of about 50 members (research engineers, predocs, postdocs, senior scientists), the AC group aims at better understanding and predicting the spatiotemporal variations of atmospheric pollutants along with their effects upon air quality, weather and climate (see a video presentation of the group in). This is addressed through the continuous development and application of numerical models over multiple scales, from weather to climate and from global to urban scales. The AC group is the research backbone of the Multiscale Online Non-hydrostatic AtmospheRe CHemistry model (MONARCH), a cutting-edge atmospheric composition model used for both research and operational activities, that contains advanced chemistry and aerosol packages coupled online with a meteorological driver. MONARCH is part of the ensemble Copernicus Atmospheric Monitoring Service (CAMS) regional air quality forecasting system that provides operational forecast and analysis over Europe. CAMS is a key component of the European Union’s Earth observation system. MONARCH also runs operationally at the first World Meteorological Organization (WMO) Barcelona Dust Regional Center (BDRC) for Northern Africa, the Middle East and Europe, and the International Cooperative for Aerosol Prediction (ICAP) ensemble of global aerosol forecasts.

Exploitation of novel high-resolution emission inventories requires going toward higher spatial resolution to prepare the next generation of air quality predictions in Europe. We are seeking a highly motivated Research Engineer or Postdoctoral Researcher to join our team in the development of a physically consistent super-resolution downscaling deep learning models for chemical transport models (CTMs), able to predict high-resolution atmospheric composition fields (1-3 km) from coarse regional simulations (10-20 km). He / she will explore, adapt and improve cutting-edge deep learning architectures proposed in the field of computer vision and weather / climate sciences, including image- and video-based convolutional neural networks. Different strategies will be investigated to improve the physical consistency of the predicted high-resolution concentration fields, using soft and / or hard constraints. More sophisticated types of machine learning including for instance graph neural networks will also be explored. This research will start focusing on the Iberian Peninsula but should then scale to European scale (and in the future, to global scale).

The successful applicant will be part of an active group of researchers focusing on leveraging deep learning technologies to address key scientific and policy-oriented challenges (currently about 7 people in the AC group, about 15-20 in the BSC Earth Sciences department). To conduct this research, he / she will have access to the groundbreaking High-Performance Computing infrastructure of BSC, notably MareNostrum 5, one of the most powerful supercomputers in Europe, with a peak performance of 314 Pflops, 200 PB of storage and 400 PB of active archive, and an accelerate partition including 1120 nodes composed of 4 NVIDIA GPUs nodes. The candidate will also benefit from the collaboration with the Computational Earth Sciences group of the department, composed of experts in HPC, software development, and AI.

This activity is part of a large initiative on the “Modernization of observation networks and digitalization of production processes for the development of intelligent meteorological services in the context of climate change” in the framework of the European Recovery, Transformation, and Resilience Plan funded by the European Union-Next Generation EU.

Key Duties

• Develop and implement deep learning architectures CNNs, GANs, diffusion models) for spatial super-resolution of atmospheric composition fields generated by atmospheric chemistry models
• Train and validate models using historical high-resolution observational datasets and CTM outputs
• Integrate physical constraints and uncertainty estimation within the machine learning workflow
• Collaborate closely with atmospheric scientists, and participate to the intellectual life of the group
• Present model developments and research findings, contribute to scientific publications, and other duties as assigned

Requirements

• Education

PhD (or MSc with strong experience) in computer science, data sciences, Earth sciences, applied mathematics, physics, or related discipline

• Essential Knowledge and Professional Experience

Strong background in deep learning, especially in image super-resolution or geospatial applications

• Experience with ML frameworks PyTorch, TensorFlow) and high-performance computing environments
• Demonstrated expertise in designing and implementing machine learning models from scratch
• Excellent programming skills in Python
• Additional Knowledge and Professional Experience

Experience working in HPC environment (including bash)

• Experience in Earth sciences will be valued
• Experience with graph neural networks will also be valued
• Experience with revision control systems SVN or Git)
• Competences

Very good interpersonal skills

• Fluency in English
• Excellent written and verbal communication skills
• Ability to take initiative, prioritize and work under set deadlines
• Ability to work both independently and within a team

Conditions

• The position will be located at BSC within the Earth Sciences Department
• We offer a full-time contract a good working environment, a highly stimulating environment with state-of-the-art infrastructure, flexible working hours, extensive training plan, restaurant tickets, private health insurance, support to the relocation procedures
• Duration : Open-ended contract due to technical and scientific activities linked to the project and budget duration
• Holidays : 23 paid vacation days plus 24th and 31st of December per our collective agreement
• Salary : we offer a competitive salary commensurate with the qualifications and experience of the candidate and according to the cost of living in Barcelona

J-18808-Ljbffr