Community jobs in United States

Organisation/Company IMT Atlantique Research Field Computer science Environmental science Researcher Profile First Stage Researcher (R1) Positions PhD Positions Country France Application Deadline 4 Jul 2025 - 17:00 (Europe/Paris) Type of Contract Temporary Job Status Full-time Hours Per Week 37 Offer Starting Date 1 Sep 2025 Is the job funded through the EU Research Framework Programme? Horizon Europe – COFUND Marie Curie Grant Agreement Number 101126644 Is the Job related to staff position within a Research Infrastructure? No

The PhD position is offered under an industrial track (2 years at IMT Atlantique + 9 months at Ifremer, France and 3 months at an international academic partner Delft University of Technology, The Netherlands).

1.1. Domain and scientific/technical context

Satellite remote sensing observations provide a variety of measurements of oceanic and atmospheric variables at a rich spatial variability. These data offer a broad perspective on environmental phenomena, including sea level variations, surface wind fields, and ocean surface roughness. For instance, SAR wave mode (WV) captures 20x20 km image patches at 5 m pixel resolution and is unaffected by cloud cover or time-of-day. This makes it a critical source of information for monitoring the ocean. SAR data provide highly accurate direct observations of different oceanic phenomena, including internal waves and mesoscale eddies, which in turn offer valuable insights into ocean dynamics and energy transfer mechanisms between the ocean and the atmosphere.

Artificial Intelligence (AI) technologies and models open new paradigms for analyzing ocean remote sensing data. However, unlike regular images where correlations are typically local—favoring the use of standard convolutional neural networks (CNNs)—SAR images, along with other ocean remote sensing observations, can exhibit very longrange correlations. This characteristic challenges the application of standard CNN-based vision models. Additionally, while Vision Transformer (ViT) models [2] offer a different approach, they also struggle to capture these long-range correlations due to the quadratic scaling of attention mechanisms.

Recent methods based on state-space models [3] have demonstrated strong capabilities in modeling very long sequences. In this context, these methods provide the perfect alternative to standard deep learning approaches in representing long correlation in physical observations. This project aims to leverage these state-space representations to develop foundation models of ocean satellite data observations. These models are expected to better capture long-range dependencies more effectively and to improve the detection and classification of physical phenomena.

1.2. Scientific/technical challenges

The primary challenge addressed in this research is the ability to define deep learning models able of modelling long-range dependencies in ocean remote sensing data. Specifically, this study will focus on:

• Identifying the limitations of existing deep learning models in capturing long-range correlations in SAR images and multimodal ocean data.
• Developing new foundation models based on state-space representations to improve feature extraction from satellite data.
• Evaluating the effectiveness of these models in various applications including classification of physical phenomena and inverse problems.

1.3. Considered methods, targeted results and impacts

The research will be divided into two parts:

• Pretraining the state-space based foundation model using self-supervised learning techniques on largescale SAR and multimodal ocean datasets [2].
• Fine-tuning on labelled SAR data images [4].

Model performance will be compared against traditional ViTs and CNNs in terms of accuracy, computational efficiency, and robustness in detecting oceanic phenomena.

Targeted Results:

• We expect this project to lead to a new foundation model that can better capture long-range dependencies in very high-resolution satellite data. This model will allow better classification of the various ocean phenomena that are captured.
• A benchmark dataset and evaluation framework will also be developed for future studies.

Impact: The project will have significant implications on multiple fronts:

• On the Sentinel-1 mission: This project provides new means to better exploiting SAR observations, further improving the impact of the Sentinel-1 mission and other satellite missions that will be used in this study.
• On the scientific community: This project aims at delivering a novel tool that facilitates the processing and analysis of large-scale satellite ocean data. It also allows to transferer state-of-the-art technological developments in AI architectures to actors and researchers in the environmental science.

1.4. Interdisciplinarity aspects

This project explores the integration of state-of-the-art AI models to improve the representation of physical phenomena observed through remote sensing. It is a highly interdisciplinary initiative that brings together research from AI, machine learning, remote sensing, and oceanography to tackle the challenges of capturing and interpreting complex geophysical processes.

1.5. References

[1] Torres, R., Snoeij, P., Geudtner, D., Bibby, D., Davidson, M., Attema, E., … & Rostan, F. (2012). GMES Sentinel-1 mission. Remote sensing of environment, 120, 9-24. [2] Glaser, Y., Stopa, J. E., Wolniewicz, L. M., Foster, R., Vandemark, D., Mouche, A., … & Sadowski, P. (2024).

[2] WV-Net: A foundation model for SAR WV-mode satellite imagery trained using contrastive self-supervised learning on 10 million images. arXiv preprint arXiv:2406.18765.

[3] Zhu, L., Liao, B., Zhang, Q., Wang, X., Liu, W., & Wang, X. Vision mamba: Efficient visual representation learning with bidirectional state space model. arXiv 2024. arXiv preprint arXiv:2401.09417.

[4] Wang, C., Stopa, J. E., Vandemark, D., Foster, R., Ayet, A., Mouche, A., … & Sadowski, P. (2025). A multi-tagged SAR ocean image dataset identifying atmospheric boundary layer structure in winter tradewind conditions. Geoscience Data Journal, 12(1), 1-14.

The PhD student will stay 9 months at Ifremer.

• Academic international partner: The PhD thesis includes a 3-months stay at an international academic partner, probably at Delft University of Technology, The Netherlands.

2.2. Hosting organizations

IMT Atlantique , internationally recognized for the quality of its research, is a leading French technological university under the supervision of the Ministry of Industry and Digital Technology. IMT Atlantique maintains privileged relationships with major national and international industrial partners, as well as with a dense network of SMEs, start-ups, and innovation networks. With 290 permanent staff, 2,200 students, including 300 doctoral students, IMT Atlantique produces 1,000 publications each year and raises 18€ million in research funds.

2.2.2. Ifremer

Ifremer is the French research institute entirely dedicated to gathering knowledge of the ocean. Through its scientific and technological research, innovations and expertise, Ifremer contributes to protecting and restoring the ocean, sustainably managing marine resources and environments, and sharing marine data and information. Ifremer is involved in scientific initiatives and programs of national, European and international scope.

• AI and Deep Learning Expertise: Strong knowledge of AI techniques, especially deep learning models, and their applications to large datasets.
• Familiarity with Large Language Models (LLMs): Understanding of how LLMs work and their potential application in analyzing large-scale remote sensing data.
• Mathematical and Statistical Modelling: Proficiency in mathematical modeling, including statistical techniques and methods for solving complex inverse problems. Knowledge of state-space models will be appreciated.
• Research Enthusiasm and Curiosity: A strong interest in interdisciplinary research, with a willingness to explore new approaches at the intersection of AI, and remote sensing.
• Ability to critically analyze data, generate hypotheses, and develop novel methodologies, coupled with strong problem-solving skills.

Research Field Computer scienceEnvironmental science

APhD programme of high quality training : 4 reasons to apply

• SEED is a programme of excellence that is aware of its responsibilities: to provide a programme of high quality training to develop conscientious researchers, including training in responsible research and ethics.
• SEED’s unique approach of providing interdisciplinary, international and cross-sector experience is tailored to work in a career-focused manner to enhance employability and market integration.
• SEED offers a competitive funding scheme, aiming for an average monthly salary of EUR 2,000 net per ESR, topped by additional mobility allowances as well as optional family allowances.
• SEED is a forward-looking programme that actively engages with current issues and challenges, providing research opportunities addressing industrial and academic relevant themes.

Eligibility criteria. In accordance with MSCA rules, SEED will open to applicants without any conditions of nationality nor age criteria. SEED applies the MSCA mobility standards and necessary background. Eligible candidates must fulfil the following criteria

• Mobility rule: Candidates must show transnational mobility by having not resided or carried out their main activity (work, studies, etc.) in France for more than 12 months in the three years immediately before the deadline of the co-funded program's call (March 20 for call #3). Compulsory national service, short stays such as holidays and time spent as part of a procedure for obtaining refugee status under the Geneva Convention are not taken into account.
• Early-stage researchers (ESR): Candidates must have a master’s degree or an equivalent diploma at the time of their enrolment and must be in the first four years (full-time equivalent research experience) of their research career. Moreover, they must not have been awarded a doctoral degree.
  
  Extensions may be granted (under certain conditions) for maternity leave, paternity leave, as well as long-term illness or national service.