At Angely, I fine-tuned a small object detection CNN that runs 2.5 times faster than the main system. I also increased tensor processing speed by 15% and redesigned the vision module’s communication protocol, making it 30% faster.
At EmTech Space, I implemented an Earth Observation preprocessing pipeline for Simera Sense HyperScape100 imagery, cutting processing time by an average of 20% through parallel processing. I also served as technical manager and backend engineer on the Golden Twins platform, building report and map generation to meet runtime targets.
As an independent researcher, I designed SigmML, a recurrent metric meta-learning network on the symmetric positive definite matrix manifold. My research outperformed state-of-the-art models in two papers, reducing equal error rate from an average of 0.3% to under 0.05% in intra- and cross-dataset tests.
I’m a PhD student in Electrical and Electronics Engineering at the University of West Attica, researching synthetic data generation with computer vision and data-driven models to improve the robustness of systems with limited training data. I’m also a first author of three peer-reviewed papers published at WACV, ICDAR, and Expert Systems with Applications.
