Hi! My name is Mateo Espinosa ZarlengaFollowing Spanish naming customs, my last name has two words in it (i.e., ‘Espinosa Zarlenga’). and I, at least last time I checked, look approximately like the dude in the picture somewhere near this text. Currently, I am a Junior Research Fellow (JRF) in Computer Science at Trinity College, Oxford. I did my PhD on interpretable AI at the University of Cambridge under the wonderful supervision of Professor Mateja Jamnik and Dr Zohreh Shams. In my research, I am generally interested in the overlap of AI interpretability, representation learning, model control, and human-AI interaction. In particular, I am interested in understanding how deep learning models can be steered, helped, and controlled by users through manipulations of representations that are aligned with human-understandable concepts. If this sounds possibly interesting for you, please take a look at my research page for a somewhat updated compilation of the work I am doing these days.
Before crossing the pond for my MPhil and PhD, I worked for three and a half years at Cerebras Systems, a machine learning hardware accelerator startup in Silicon Valley. During this time, I probably ate my life quota of sourdough while working on some exciting projects at the intersection of software and hardware for AI. Before that, I lived for four wonderful years in Ithaca, NY where I was a Master of Engineering student (2016-2017) and an undergraduate student (2013-2016) at Cornell University. Keep in mind that Ithaca’s namesake completely ignores the weather of its Homeric counterpart. So most of that time was spent happily buried under snow. Before my student life, I was in the gorgeous and oxygen-deprived city of Quito, Ecuador, where I was born and raised.
On this site I will be posting random thoughts and projects I get to work on in my free time (or at least that’s the idea). In the meantime, if you want to contact me or share some ideas with me (I am always open to collaborations!), please do not hesitate to do so.
In case you were wondering about the name of this site: the hairy ball theorem says that you cannot comb a hairy ball flat. However you comb it, at least one cowlick is left over, a point where the hair has nowhere to go but straight up. You are welcome to check this yourself below.
A small disclaimer: unfortunately, given how things work today, there’s a need to say that everything on this site is highly subjective and one hundred percent my personal opinion, not that of my employers/supervisors or any institution I am affiliated with.