This user has not added any information to their profile yet.
I obtained my PhD in Computer Science from the University of Lille in 2018, and have been a research scientist in the NLP group at NAVER LABS Europe since then.
My PhD work was about neural machine translation, speech translation, and automatic post-editing.
In the past few years, I have worked mostly on machine translation topics: how to make NMT models more robust to input noise (e.g., noisy user-generated content like restaurant reviews); domain adaptation; and document-level machine translation. We participated in several shared tasks about these topics (WMT19 robustness, WNGT19 doc-level generation and translation, WMT20 robustness/chat/biomedical).
More recently, I have focused on multilingual NMT: improving zero-shot translation; speeding up MNMT inference; efficiently adding new languages to MNMT (aka “continual learning”), sometimes even without parallel data (“unsupervised MNMT”); and multilingual multi-domain NMT.