Published multiple papers such as 'KL-Regularized RLHF with Multiple Reference Models: Exact Solutions and Sample Complexity' (NeurIPS 2025), 'Large Language Models can Become Strong Self-Detoxifiers' (ICLR 2025), etc.; holds several patents including 'Wasserstein Barycenter Model Ensembling' (US 12254390).
Research Experience
Served as a Principal Research Scientist at IBM since 2015.
Education
Received his PhD in Computer Science from MIT, CSAIL in February 2015, under the supervision of Professor Tomaso Poggio; obtained an engineering diploma from Ecole Polytechnique Paris France and a master's degree in Applied Mathematics from Ecole des Mines de Paris in 2011.
Background
Research interests include Deep Learning, Machine Learning, Optimal transport, multimodal learning, Statistical Learning Theory, Computer Vision, and Artificial Intelligence. He has been a Principal Research Scientist at IBM since April 2015.
Miscellany
Top collaborators include Mattia Rigotti, Payel Das, Pin-Yu Chen, and others.