Richard He Bai
I’m a Research Scientist at Apple MLR working on self-improving LLMs.
My broader interest is in how models can learn from their own outputs and feedback, and interact with people in real time. I’ve developed methods for data-efficient pretraining (corpus rephrasing), LLM post-training (self-distillation and self-training from test-time scaling), and real-time speech agents (speech tokenization, streaming speech generation, and on-device conversation).
I received my PhD in Computer Science from UWaterloo, advised by Ming Li, and my Master’s from UCAS, advised by Chengqing Zong.
