HPC workflows, climate science, and machine learning

Aina Gaya Àvila

Research engineer and PhD candidate working across Barcelona and Munich on large-scale climate workflows, Earth observation, and machine learning for geosciences.

Get in touch Explore my work GitHub LinkedIn

Barcelona Supercomputing Center Technical University of Munich HPC, climate, and AI

Current profile

Operational workflows for climate simulation at scale

I build the infrastructure around computational Earth science: workflow automation, portability across supercomputers, data pipelines, and ML-ready climate datasets.

HPC workflows Bash, Python, YAML, containers, and workflow orchestration for operational climate simulations on EuroHPC systems.

Earth observation and AI PhD work on coastal sea surface height with SWOT altimetry and deep learning.

Leadership Team coordination, user support, documentation, debugging, and delivery under research and operational deadlines.

Experience

Roles across HPC engineering, climate workflows, and research.

The last few years have centered on research engineering for Destination Earth and related projects, alongside an ongoing PhD in Earth observation and machine learning.

01/2026-Present

PhD Candidate

Technical University of Munich, Data Science in Earth Observation

Researching coastal sea surface height with novel SWOT satellite altimetry data and deep learning methods, with a toolkit that includes PyTorch, GPUs, CUDA, and Earth observation workflows.

02/2023-Present

HPC Workflows Engineer

Barcelona Supercomputing Center, Computational Earth Sciences

Developing and coordinating operational workflows for DestinE projects including ClimateDT, EERIE, DE-393, and EDITO-Model Lab. Since December 2024, leading a team of 6 engineers and the workflow activity within a DG-CONNECT Destination Earth contract. Built workflows for kilometer-scale climate simulations, containerized NEMO runs, automated benchmarking, and ML-ready climate dataset creation. Daily work includes portability across MareNostrum5 and LUMI, plus occasional deployments on Leonardo, JUPITER, ATOS, Levante, JUWELS, and related systems.

11/2022-06/2023

Undergraduate Researcher

Condensed Matter Physics Department, Universitat de Barcelona

Final degree project with Dr. Alberto Fernández-Nieves's group on quantifying the colloidal gelation transition using image compression, with a focus on Python-based data processing and scientific writing.

09/2021-08/2022

Undergraduate Researcher

UBICS, Institute of Complex Systems

Developed a computational application of a three-state opinion model under the supervision of Dr. Albert Díaz-Guilera, working with Python, NetworkX, NumPy, SciPy, Matplotlib, Pandas, Linux, and Bash.

2015-2022

Private Teacher

Independent

Taught maths, physics, chemistry, and English in one-to-one settings, building a strong habit of clear explanation and adapting technical material to different audiences.

Education and activities

Physics training expanded into computational modelling and climate AI.

My academic path moves from fundamental physics into multiscale computational modelling, uncertainty quantification, and now Earth observation research at doctoral level.

MSc in Atomistic and Multiscale Computational Modelling in Physics

Completed jointly at Universitat Politècnica de Catalunya and Universitat de Barcelona from September 2023 to June 2025. My thesis focused on AI-assisted uncertainty quantification in high-resolution climate projections, using AIFS and NeuralGCM on MareNostrum5.

BSc in Fundamental Physics

Earned a physics degree at Universitat de Barcelona with honors in Computational Physics, Image Processing and Computer Vision, and Theoretical Mechanics.

Skills

Languages, platforms, and tools I use daily.

My toolkit spans workflow automation, scientific programming, containerized execution, data handling, and collaboration around large operational research systems.

Programming and technical tools

Python Fortran Bash LaTeX Gnuplot JavaScript HTML NumPy Pandas SciPy scikit-learn PyTorch Linux Git GitHub Actions Docker Apptainer Autosubmit SLURM NetCDF GRIB

Platforms and domains

High Performance Computing Climate workflows Portability Containers CI/CD Machine Learning Deep Learning Satellite data Altimetry OpenMP/MPI

Languages

Catalan (native) Spanish (native) English (fluent) French (beginner)

Strengths

Leadership Problem-solving Project management User support Scientific writing Open source development Research collaboration

Publications and talks

Publications and presentations

Publications

Complex Networks

Ferri, I., Gaya-Àvila, A., Díaz-Guilera, A., Three-state opinion model with mobile agents. Chaos 33, 093121 (2023). doi.org/10.1063/5.0152674

Earth and Space Science Informatics

John, A., Beyer, S., (...) Gaya-Àvila, A., et al. Global Storyline Simulations at the Kilometre-scale. ESS Open Archive (2024). doi.org/10.22541/essoar.173160166.64258929/v1
Doblas-Reyes, F. J., Kontkanen, J., Sandu, I., Acosta, M., (...) Gaya-Àvila, A., et al. The Destination Earth digital twin for climate change adaptation. EGUsphere (2025). doi.org/10.5194/egusphere-2025-2198

Conference Papers and Presentations

European Geosciences Union General Assembly, 04/2024

Gaya-Àvila, A. et al., A workflow for the Climate Digital Twin, EGU General Assembly 2024, Vienna, Austria, 14-19 Apr 2024, EGU24-2533. doi.org/10.5194/egusphere-egu24-2533
Roura-Adserias, F., Gaya-Àvila, A. et al., The data streaming in the Climate Adaptation Digital Twin: a fundamental piece to transform climate data into climate information, EGU General Assembly 2024, Vienna, Austria, 14-19 Apr 2024, EGU24-2164, 2024. doi.org/10.5194/egusphere-egu24-2164
Keller, K., Acosta, M., Ghosh, S., Gaya Avila, A., Wagner, I., and Paronuzzi, S., The Backbone of the Destination Earth Climate Adaptation Digital Twin, EGU General Assembly 2024, Vienna, Austria, 14-19 Apr 2024, EGU24-9492, 2024. doi.org/10.5194/egusphere-egu24-9492

European Geosciences Union General Assembly, 04/2025

Gaya-Àvila, A., de Paula Kinoshita, B., Paronuzzi Ticco, S. V., Tintó Prims, O., and Castrillo, M., A workflow for cloud-based and HPC simulations with the NEMO ocean model using containers, EGU General Assembly 2025, Vienna, Austria, 27 Apr-2 May 2025, EGU25-1511, 2025. doi.org/10.5194/egusphere-egu25-1511
Roura-Adserias, F., Gaya-Avila, A., Arriola i Meikle, L., Gonzalez-Yeregi, I., De Paula Kinoshita, B., Tollander de Balsch, J., and Castrillo, M., ClimateDT Workflow: A containerized climate workflow, EGU General Assembly 2025, Vienna, Austria, 27 Apr-2 May 2025, EGU25-4466, 2025. doi.org/10.5194/egusphere-egu25-4466
Garcia Lopez, A., Arriola Meikle, L., Montane Pinto, G., Castrillo, M., de Paula Kinoshita, B., Ferrer Escuin, E., and Gaya-Àvila, A., Enabling reliable workflow development with an advanced Testing Suite, EGU General Assembly 2025, Vienna, Austria, 27 Apr-2 May 2025, EGU25-8305, 2025. doi.org/10.5194/egusphere-egu25-8305
Gonzalez-Yeregi, I., Bretonnière, P.-A., Gaya-Àvila, A., and Roura-Adserias, F., Generic State Vector: streaming and accessing high resolution climate data from models to end users, EGU General Assembly 2025, Vienna, Austria, 27 Apr-2 May 2025, EGU25-4355, 2025. doi.org/10.5194/egusphere-egu25-4355
Caprioli, S., von Hardenberg, J., Nurisso, M., Davini, P., Nazarova, N., Ghosh, S., Cadau, M., Mubarak Rajput, M., Gaya-Àvila, A., and Zimmermann, J., Model evaluation for km-scale simulations within the Climate Adaptation Digital Twin: the AQUA approach, EGU General Assembly 2025, Vienna, Austria, 27 Apr-2 May 2025, EGU25-10347, 2025. doi.org/10.5194/egusphere-egu25-10347

NeurIPS, 12/2025

Gaya-Àvila, A., Mozaffari, A., Duarte, A., Tintó Prims, O., A modular framework to run AI-based models from high-resolution climate projections, Tackling Climate Change with Machine Learning, San Diego, USA, 7 Dec 2025. neurips.cc/virtual/2025/loc/san-diego/126907