Ching Fang

Hi, I'm Ching!

I am a research scientist at Goodfire AI, where I work on using AI interpretability to understand scientific models and guide scientific discovery. I'm interested in many topics in AI and neuroscience! My past research has involved applications of mechanistic interpretability (for instance, in LLMs that use encoded reasoning or to understand in-context reinforcement learning). I'm also interested in understanding how representation geometry influences learning (for instance, in deep RL networks as a model of learning in the brain).

I was previously a postdoctoral researcher at Harvard University. Before that, I completed my PhD at Columbia University's Center for Theoretical Neuroscience, where I studied models of episodic memory in the hippocampus and key-value memory. I was also a ML research intern at Apple, working on multimodal foundation models for biosignals. I completed my undergraduate degree in computer science and molecular biology at UC Berkeley, where I studied learning in brain-machine interfaces.

Updates

Search