jgcarrasco

Welcome to my homepage!

I’m currently finishing my PhD studies at the University of Alicante. My research is mostly focused on trying to understand the inner workings of neural networks, mechanistic interpretability and its applications. Before getting into deep dearning/ML I got a physics degree. During that time I discovered ML and loved the combination of programming/math/problem solving abilities.

You can also find me at GitHub, X, LinkedIn, Google Scholar.

projects

• Unsupervised Ink Detection with DINO: Work done for the Vesuvius Challenge to detect ink from carbonized scrolls. Currently, we have a lot of unlabeled data (terabytes of x-ray scan data from scrolls) and very little labeled data (manually annotated ink), so the idea is to develop AI techniques to take advantage of these large amounts of unlabeled data.

• baccpropagation: Simple neural network implemented in C