Wafer builds autonomous AI agents that act as performance engineers, profiling, diagnosing, and optimizing GPU inference across the entire stack from kernels to models to production pipelines. The company provides serverless and dedicated inference infrastructure for open-source LLMs, achieving significant speed improvements through kernel optimization and serving-stack rewriting. It is already working with major chip and cloud players to optimize code for custom silicon. Founded in 2025 by two University of Chicago grads, Wafer was part of YC's Summer 2025 batch.