Ivar van der Spoel

Information

Bio

I am Ivar van der Spoel, a Computer Science student with a strong foundation in Advanced Computing and Systems and a particular interest in Artificial Intelligence. My academic journey began at Leiden University in 2021, where I completed my Bachelor’s degree in Computer Science in 2024. I am currently working toward the completion of my Master’s thesis.

Current Role

At present, I am a Master’s student at Leiden University, completing my thesis in collaboration with ESA Φ-lab. As a visiting researcher, I investigate robustness in Neural Architecture Search for Earth observation. A central challenge in this work is the domain gap between training data and real-world data, which is further complicated by natural degradations that affect the quality of acquired observations over time. My research begins by analyzing and benchmarking the state of the art through the application of realistic degradations to simulated data. The broader goal is to use multi-objective optimization to develop reliable and robust models for Earth observation, with a focus on practical deployment in challenging real-world conditions.

Areas of Expertise

My expertise spans several areas of Computer Science, ranging from low-level topics such as computer architecture, operating systems, and CUDA programming to higher-level areas including distributed and cloud systems. In addition, I have a strong interest in Artificial Intelligence, particularly in Natural Language Processing, Deep Learning, and Automated Machine Learning. These combined interests and skills enable me to contribute to ESA Φ-lab CIN’s mission by investigating robustness in Neural Architecture Search for Earth observation. In particular, I aim to support the development of models that remain effective under data shifts and degradations, helping make Earth observation systems more dependable and applicable in practice.

Vision for the Future

I am driven by a vision of a future in which compact, efficient, and robust neural networks can be reliably deployed on edge devices for Earth observation. My work aims to advance knowledge on the development of small neural networks suitable for on-edge deployment, while addressing the domain gap between simulated and acquired satellite data, as well as the natural degradations that arise in real-world settings. By improving the robustness and reliability of these models, I hope to contribute to Earth observation systems that are more adaptive, trustworthy, and useful in practice. In the long term, I believe this research can help enable more resilient AI solutions for space and environmental applications, where efficiency, accuracy, and robustness are all essential. shifts in mind.

Project as CIN Researcher

Coming Soon