I’m a Computer Science PhD from University of Buenos Aires, Argentina. I’m passionate about Machine Learning and Software Engineering, two topics which I’ve studied in-depth during my PhD.

Right now, I’m fascinated by the way AI models are becoming incredibly common and making a big impact in society, even though we still don’t fully understand how they work. I believe that in order to ensure that AI is safe, we need to achieve a comprehensive understanding of the inner workings of these models. This has motivated me to dive deep into the field of Mechanistic Interpretability, taking part in the ML Alignment & Theory Scholars (MATS) program, where I'm working on a project to build a benchmark of semi-synthetic but realistic transformers with known circuits for evaluating mechanistic interpretability techniques.

Contact Information E-Mail: iarcuschin@dc.uba.ar