Ammar Kheder

Doctoral Researcher in Computational Engineering · LUT University · AMC-Lahti

Teaching neural networks how the atmosphere flows.

Ammar Kheder Ammar Kheder, portrait illustrated by Mitsuhiro Arita

I develop neural network architectures that explicitly encode atmospheric physics, like terrain-atmosphere interactions and advective transport, to push spatial resolution and forecast accuracy for air quality and Earth system prediction. Based at LUT University within the Atmospheric Modelling Centre (AMC-Lahti), supervised by Prof. Michael Boy and Assoc. Prof. Zhi-Song Liu.

I have hands-on experience running large-scale distributed training on the LUMI supercomputer, scaling experiments up to 1,024 AMD MI250X GPUs for training vision transformer models on high-resolution atmospheric reanalysis data.

MSc in Engineering (Big Data & AI) from EiCnam Paris, with a one-year apprenticeship at INRIA Bordeaux within the Mnemosyne team, co-led by Frédéric Alexandre and Nicolas Rougier.

Keywords. Air quality, atmospheric modelling, physics-informed neural networks, vision transformers, climate, spatial downscaling, high-performance computing, large-scale distributed training.

Download CV (PDF) ↓

npj

Nature Portfolio
1 publication, 2026

10

Citations
across 4 works

1,024

GPUs scaled
LUMI / AMD MI250X

TopoFlow vs CAQRA over the Sichuan Basin: maps, transect and conceptual diagram

Latest paper · April 2026 · lead author

TopoFlow

Topography-aware pollutant flow learning for high-resolution air quality prediction. A physics-informed Vision Transformer with wind-following patches and elevation-aware attention.

npj Climate and Atmospheric Science, Nature Portfolio.

Read the case study →

News

  1. Apr 2026 New TopoFlow published in npj Climate and Atmospheric Science (Nature Portfolio).
  2. May 2026 Two-day research stay (5–6 May) with AMC member Prof. Jia Chen and her group at the Technical University of Munich (TUM), marking the start of a new international collaboration.
  3. Apr 2026 Talk accepted at IAC 2026 (International Aerosol Conference), Xi’an, China.
  4. Mar 2026 CRAN-PM preprint on arXiv.
  5. Mar 2026 Inverse Neural Operator preprint on arXiv.
  6. Feb 2026 TopoFlow preprint on arXiv.
  7. Jan 2026 Invited visit to my former programme BUT SD Niort (5 January) to share my academic and professional journey with current students.

All news in the archive →

Publications

  1. TopoFlow: topography-aware pollutant Flow learning for high-resolution air quality prediction

    A. Kheder, H. Toropainen, W. Peng, S. Antão, J. Chen, M. Boy, Z.-S. Liu

    npj Climate and Atmospheric Science, Nature Portfolio, 2026.

    Nature arXiv Code Project

  2. CRAN-PM: Cross-Resolution Attention Network for High-Resolution PM2.5 Prediction

    A. Kheder et al.

    Preprint, 2026.

    arXiv Code Project

  3. Inverse Neural Operator for ODE Parameter Optimization

    Z.-S. Liu, W. Peng, H. Toropainen, A. Kheder, A. Rupp, H. Froning, X. Lin, M. Boy

    Preprint, 2026.

    arXiv

  4. Deep Spatio-Temporal Neural Network for Air Quality Reanalysis

    A. Kheder, B. Foreback, L. Wang, Z.-S. Liu, M. Boy

    Scandinavian Conference on Image Analysis (SCIA), Springer LNCS, 2025.

    arXiv Springer Code

Experience

2024 – nowJunior Researcher & Teaching Assistant →
LUT University, Lahti, Finland
2023 – 2026CEO, Wabel Group
AI services and web development
2021 – 2022Research Engineer (Apprenticeship)
INRIA Bordeaux, Mnemosyne team
2021 – 2023Volunteer Firefighter →
Sapeurs-pompiers des Deux-Sèvres

Education

2024 – nowPh.D. Computational Engineering
LUT University, Finland
2021 – 2024MSc Engineering, Big Data & AI
EiCnam Paris, with apprenticeship at INRIA Bordeaux
2019 – 2021Bachelor, Data Science
BUT Niort. Statistics, Big Data, Business Intelligence
2016 – 2017BIA Aeronautical Initiation →
Lycée Paul-Guérin, Niort