WORK-STUDY - DEEP LEARNING FOR THE RECONSTRUCTION OF SURFACE OCEAN CURRENTS - M/F

Description:

CLS is developing ADVANCED METHODS TO RECONSTRUCT HIGH TEMPORAL RESOLUTION SURFACE OCEAN CURRENTS FROM MULTI-SOURCE DATA (satellites, numerical models, in situ observations).

Recent work has enabled the implementation of an initial Deep Learning approach on a regional area. The objective of this work-study program is to continue and improve this method, in particular by studying its robustness and its ability to generalize to other oceanic regions with varied dynamics, such as the North Atlantic and equatorial zones.

WORK-STUDY OBJECTIVES

The apprentice will contribute to:

• Improving existing Deep Learning architectures for the spatio-temporal reconstruction of surface currents,
• Strengthening the spatial generalization capacity of the models between different oceanic regions,
• Analyzing the limits and performance of the method according to dynamic regimes,
• Proposing methodological improvements and adapted quality indicators.

MAIN MISSIONS

• Getting to grips with input data (oceanic fields, winds, in situ observations) and existing processing pipelines.
• Improving Deep Learning architectures (spatio-temporal models, attention mechanisms, multi-source data fusion).
• Extending the method to several oceanic zones (regional → North Atlantic → equatorial zones).
• Analyzing the transferability of models between regions and comparing performance.
• Implementing adapted learning strategies (regional models vs. multi-region).
• Validating results using in situ observations and physical diagnostics.
• Critical analysis of results and contribution to their promotion (reports, internal presentations).

As part of your duties, your main missions will consist of:

• Conducting a discovery literature review on general and spatial oceanography, the different components of ocean currents, and their interactions.
• Training on observations that provide information on ocean currents (drifting buoys, HF radars, altimetry).
• Improving existing products by integrating new reference observations or new methodologies.
• Performing critical analysis of results.
• Contributing to the promotion of results through scientific publications.

Profile sought:

• Student in an engineering school or Master's degree (data science, applied mathematics, oceanography, spatial).
• Good level in Python and solid foundations in Deep Learning.
• Strong interest in geophysical data analysis and oceanography.
• Ability to work autonomously and analytically.
• Good level of technical English.

We wish to integrate enthusiastic, passionate, and curious apprentices into our teams who are ready to invest themselves daily.

Open-mindedness, interpersonal skills, flexibility, autonomy, and analytical spirit are expected.

If teamwork is paramount to you, if you are proactive, and if your motivation is unwavering, then do not hesitate any longer, and join us!