진학프로석·박사 채용 접수 플랫폼

Presentation of the project :

The Previzo project aims to develop an AI-based decision-support tool to optimize water resource management, particularly by anticipating freshwater quality degradation, water-related crises, and conflicts over water use. In a context shaped by climate change, eutrophication of aquatic environments, and changes in hydrological regimes, predictive monitoring of river water quality represents a major environmental, health, and socio-economic challenge.

The project builds on multispectral remote sensing, the analysis of heterogeneous environmental data, and the development of frugal artificial intelligence approaches to better detect, estimate, and anticipate changes in water quality parameters in continental aquatic environments. Particular attention is given to the integration of knowledge derived from the physics of hydrological, optical, and biogeochemical processes, as well as to model reduction methods, in order to design approaches that are more robust, interpretable, and computationally efficient.

At the interface between artificial intelligence, Earth observation, physics-based modelling, and environmental sciences, Previzo aims to contribute to the development of operational, reliable, and explainable tools to support water management stakeholders in monitoring and anticipating risks affecting freshwater quality.

Presentation of the Service / laboratory :

The PRISME laboratory is a research unit jointly affiliated with the University of Orléans and INSA Centre Val de Loire (UR 4229). Its scientific scope and methodological approaches fall within the scientific field of Science and Technology (ST). They encompass both Engineering Sciences and Information, and Communication Sciences and Technologies.

The PRISME laboratory has a multidisciplinary vocation within the broader domain of engineering sciences and technologies, covering a wide range of disciplinary fields including combustion in engines, energy systems, explosions, aerodynamics, fluid mechanics, signal and image processing, control systems, and robotics.

Presentation of the missions :

Within the framework of the Previzo project, two postdoctoral positions are open to develop complementary approaches for the predictive monitoring of freshwater quality using multispectral remote sensing data and heterogeneous environmental data.

Topic 1: Physics-guided AI models
The first position will focus on the development of physics-guided learning models for the estimation and prediction of water quality parameters. The work will aim to integrate knowledge from hydrological, optical, and biogeochemical processes into artificial intelligence models in order to improve their robustness, interpretability, and ability to generalize across space and time. The postdoctoral researcher will develop hybrid approaches combining physical models, expert knowledge, and machine learning, and will evaluate them using multispectral and environmental data from representative case studies.

Présentation of the profile :

Applicants should hold a PhD in artificial intelligence, machine learning, computer vision, remote sensing, applied mathematics, environmental modellin. They should have strong skills in machine learning, data analysis and programming, preferably in Python, with experience using frameworks such as PyTorch, TensorFlow, scikit-learn or equivalent tools. Depending on the position, expertise in physics-guided machine learning, hybrid modelling, reduced-order modelling, model compression, knowledge distillation, remote sensing image analysis or environmental data processing will be particularly appreciated. The candidate should be autonomous, rigorous and curious, with the ability to work in a multidisciplinary environment at the interface between artificial intelligence, Earth observation, physics-based modelling and environmental sciences. Good communication skills, teamwork abilities and a capacity to contribute to scientific publications and project deliverables are expected. A good command of written and spoken English is required; knowledge of French would be appreciated but is not mandatory.