Bio

Download my CV
(last updated Oct. 2016)

I started my studies in Oriental Languages at Ca' Foscari University, Venice, in 2002. In the second year I moved to Paris to continue my bachelor at the National Institute of Oriental Languages and Civilisations (INALCO). In 2006 I graduated in Oriental Languages & International Relations (languages: Arabic and Turkish).

Somewhat by chance I came across a curious master option that was offered by INALCO: namely, Traitement Automatique des Langues. I enrolled to the master, and soon realized that NLP was a fascinating research area.

In 2008 I spent two months at the University of Trento, Italy, for an internship in spoken language understanding. During that time, I got to know the research group of Human Language Technologies at Fondazione Bruno Kessler. Soon after getting my Master from INALCO, I started to work in FBK as a research intern. My first task was to adapt a Speech Recognition system to the Arabic language.

In 2009 I officially started my PhD at FBK/University of Trento, focusing on Machine Translation under the supervision of Marcello Federico. This turned out to be the best introduction to the research world that I could ever have hoped for. In addition to my main topic ‒ improving word reordering in phrase-based SMT ‒ I had the chance to work on other interesting problems, such as morphological segmentation and domain adaptation, and to participate in a number of MT evaluation campaigns.

In summer 2011, I spent three invaluable months at Microsoft Research Redmond, working with Chris Quirk on dependency-based SMT. One year later I did another internship at Dublin City University to work on syntactic pre-ordering, and finally, in early 2013, I defended my thesis with Alex Fraser, Philipp Koehn and Christof Monz as my PhD committee. The title of my thesis is Linguistically motivated reordering modeling for phrase-based statistical machine translation.

Since September 2013, I am a post-doc researcher at the ILPS group of the University of Amsterdam, where I work in the SMT team led by Christof Monz. Currently, the main goal of my work is to improve SMT into morphologically rich languages, for instance using class-based language models or neural translation models.