At International Iberian Nanotechnology Laboratory (INL), I built the data and ML foundation for a private textile dataset combining microscopy, NIR and FTIR spectroscopy, depth maps, and structured metadata across approximately 20,000 samples.
I developed multimodal models for fabric structure, composition, GSM, and stitch density estimation. I also co-invented a patent-pending microscopy, spectroscopy, and ML system deployed in production at an industrial textile manufacturer.
At University of Coimbra – Instituto de Telecomunicações, I implemented distributed GEMM and multilayer-perceptron training and inference kernels in C for commercial UPMEM processing-in-memory DIMMs. This work produced two first-author papers and open-source implementations.
My other work includes accelerating YOLO-based colorectal-polyp detection by 1.8× and contributing a fix to the nirs4all NIR spectroscopy ML library. I’ve also co-supervised BSc and MSc students on machine-learning research projects.
