My name is Meryem M'hamdi. I recently earned my PhD in Computer Science from the University of Southern California. My research focused on advancing multilingual language models beyond traditional data-intensive, identically-distributed approaches. Specifically, I developed human-like cross-lingual transfer learning techniques using meta-learning, continual learning, and curriculum learning. These innovations enhance model generalization across low-resource and unseen languages, applied effectively in multilingual information extraction, natural language understanding, and semantic search applications.
During my journey, I collaborated with many leading multinational companies including Sony, Swisscom, Adobe Research, and more recently Microsoft.
With a rich background in computer science and representation learning and a focus on applied research, my journey has led me to Microsoft, where I currently serve as both a Data Scientist II. At the forefront of Bing's multimedia team, my efforts have been instrumental in advancing user embeddings towards more personalized recommendation systems, building from scratch a state-of-the-art large language model to refine the user experience.
As a Moroccan, I am fluent in four languages/dialects: Standard Arabic, Moroccan Arabic, French, and English, with basic proficiency in Spanish and German. I have a passion for learning new languages, which broadens my cultural understanding. I also enjoy traveling; I've visited over 10 countries and have lived in three continents, including Africa (my place of origin) and Europe, where I pursued my Master's degree. In my leisure time, I indulge in activities like yoga, zumba, and occasional singing.
If a borderless world is an impossible fantasy, reducing borders between languages at least is not out of reach. With over 6,500 spoken languages around the globe, creating a unique customer service experience for everyone is a challenge faced by...