Jiaxin Shi

I am a research scientist at Meta SuperIntelligence Labs. Previously, I was a research scientist at Google DeepMind. I work on machine learning, with an emphasis on:

Generative modeling: using probabilistic methods to capture the structure and uncertainty inherent in data
Algorithmic modeling: building models that have long-term memory, efficiently encode dependencies, and extract meaningful representations
I am particularly interested in tackling challenging inferential questions at the intersection of these two areas. These encompass topics such as variational inference, sampling, gradient estimation, and score-based modeling.