Abhinav N. Harish

I am a CS PhD student at the University of Wisconsin–Madison in Trustworthy AI advised by Prof. Grigoris Chrysos. Prior to this, I explored Reinforcement Learning (RL) for Sim2Real generalization and learning long-horizon tasks like home rearrangement.

Broadly, my research span two directions:

1. Trustworthy Generative AI. Mitigating hallucinations in diffusion models — studying the reverse-time dynamics which can result in samples drifting off-manifold and designing methods that keep generations faithful.
2. Reinforcement Learning. Sample-efficient and generalizable RL — closing the sim-to-real gap so that policies learned in simulation transfer reliably to long-horizon tasks in the real world.