Head of Inference Kernels

The Head of Inference Kernels at Etched leads a high-performance team to develop optimized kernels and inference stacks for state-of-the-art transformer models, aiming for over 10x performance enhancement compared to existing benchmarks. The role encompasses architecting best-in-class inference performance, co-designing innovative algorithmic improvements, and ensuring alignment across cross-functional teams. The ideal candidate will possess extensive experience in designing GPU kernels, a deep understanding of transformer architectures, and a demonstrated track record in managing effective engineering teams.