Sudarsh

I work on reasoning systems, inference optimization, and multimodal AI.

Currently exploring efficient reasoning architectures, latent-space methods, and real-time AI systems.

My world model is simple: build systems that expose how models think, not just what they output.