My first steps into AI were as an undergraduate in Argentina where I initially joined the local study group. Later, I was lucky enough to be selected for an exchange research program with the Real Time Computer Vision Team at the Institute of Neural Computation of the Ruhr-Universität in Bochum, Germany.
After finishing that rewarding experience, I switched fields and moved to Frankfurt to intern at the headquarters of the Deutsche Bank, in which I formally started my career in the cybersecurity world.
That internship motivated me to work in the security industry for more than six years where I had the pleasure to lead an international team in charge of raising the awareness of malicious threats and its security challenges, contributing with articles and conferences in 20+ countries.
In parallel, I explored safety concerns of genomic data, which was the topic of my master’s thesis. My goal was to better understand how FASTA files can be compromised by adversaries and affect digital—biological systems. Some of my hypothesis introduced in 2015 were later confirmed in Compromising Computers with Synthesized DNA by Key et al. 2017.
I soon realized that my primary interests lie in doing interdisciplinary research. Hence, I decided to pursue graduate school to spend more time learning about Trustworthy AI, this fascinating field in the nexus of machine learning and computer security.
Therefore, I did my PhD at the Research Institute CODE of the Universität der Bundeswehr in Munich under the supervision of Gabrijela Dreo Rodosek and Lorenzo Cavallaro from University College London, UK. My work focused on adversarial attacks to understand how ML models behave when confronted with carefully manipulated input objects. Moreover, to evaluate how the resilience of these models can be improved against adaptive adversaries.
The book of my dissertation is available at Amazon and the software package can be installed using pip.
During my PhD, I also squeezed in two exciting AI Residencies at Google X both in Mountain View and Munich, in which I worked on creating new solutions and improving their performance for real-world challenging problems.
Currently, I work on Machine Learning at SandboxAQ. Also, I am the author of the FAME library whose goal is to assess ML-based malware classifiers against adversarial attacks.
It is sometimes jokingly said that nobody knows where I am from. That is partially true because during my life I have lived in lots of places across five countries: Brasil, Uruguay, Argentina, Germany, and the United States. Despite all of them being quite different in multiple aspects, you would be surprised how similar people are regardless of where they come from.
In my free time, when away of screens, I am an active traveler (visited 50+ countries) and a curious photographer who is always excited about the next moonshot project! 💡🚀