Artificial intelligence (AI) is changing how we extract information from Earth observation data. Satellite constellations and unmanned aerial vehicles (UAVs) now collect imagery at a volume, frequency and spatial detail that traditional analysis methods struggle to handle. AI methods turn these data into timely and accurate information on vegetation, ecosystems and agricultural systems. This capability is especially valuable in arid regions such as Saudi Arabia, where ground observations are sparse and environmental change is rapid. The field is advancing on three connected fronts. Deep learning models (i.e., convolutional networks and vision transformers) are now state of the art for many tasks, including image classification, segmentation, object detection and the estimation of biophysical variables. Geospatial foundation models, pretrained on large image archives often drawn from several sensors, can be adapted to new tasks and regions with fewer labelled samples. More recently, applications built on large language models (LLMs) have started to link imagery with natural language, so users can query, analyse and interpret remote sensing data through conversational and agent-based workflows. 

This internship is hosted by the Hydrology, Agriculture and Land Observation (HALO) group at KAUST, which uses UAV and satellite data to monitor natural and agricultural ecosystems across the Kingdom. The visiting student will support the design, training, evaluation and deployment of AI methods across ongoing group projects. Depending on the student's background and project needs, the work may span the full chain from data preparation to model development and application. The student will gain hands-on experience with large Earth observation datasets, KAUST high-performance computing resources, and field-collected reference data.

Keywords for web search
Remote sensing; Artificial intelligence; Deep learning; Geospatial foundation models; Large language models; UAV; Satellite imagery; Earth observation.

• To work under the biological and environmental science and engineering department (BESE) to support HALO projects on AI applied to remote sensing, including data curation, model development, training and evaluation for UAV and satellite datasets.

• Development and benchmarking of deep learning models for tasks such as classification, segmentation, object detection and estimation of biophysical variables.