NaHyeon Park

I am a PhD student at KAIST AI, advised by Hyunjung Shim. I am currently visiting NUS, where I work with Anji Liu on reasoning in diffusion language models. My research so far has focused on making text-to-image generative models more adaptable and controllable, mainly through data-efficient training & inference-time methods.

More broadly, I aim to understand how generative models work internally: how they represent concepts, how conditioning guides generation, and how these insights can help build more interpretable, reliable, and safe AI systems.

I am always happy to chat about fun research ideas & potential collaborations! 😊