At Geograkom, I developed algorithms to process and classify LiDAR point clouds in road scenes, working with files containing up to 100 million points. I investigated data characteristics and failure modes, combining geometric processing with machine learning experiments when off-the-shelf approaches didn’t meet the task requirements.
I built a complete workflow for industrial-object classification, from extracting database data and organising labels to training a Siamese neural network and applying it to classification. I handled problem exploration, data preparation, experimentation, and quality assessment while balancing accuracy goals with practical data and compute limits.
I have an M.Sc. in Applied Mathematics from MIET. My work draws on computer vision, predictive modelling, and a mathematically grounded approach to experimental machine learning.
Japan
Spain