Defended PhD thesis at MIT and joined Meta Superintelligence Labs

I successfully defended my PhD thesis titled “Score Estimation for Generative Modeling” in the Department of Electrical Engineering and Computer Science at MIT, advised by Professor Gregory W. Wornell. The thesis addresses three primary areas within score-based and diffusion generative models: improved score estimation techniques, Bayesian-inspired frameworks for solving inverse problems such as signal separation, and efficient one-step generation via mixture distribution score estimation.

Following my defense, I joined Meta Superintelligence Labs (MSL) as a Research Scientist, where I focus on speech synthesis from Meta’s multimodal models, few-step generative modeling, neural codec training, and multimodal LLM post-training.