Staff SWE, ML Compiler Performance Engineer Co-Design at Google

Google is hiring a Software Engineer, TPU Performance to define and own the strategy for enabling open-source machine-learning models to achieve TPU performance before silicon pilot. The role involves adapting models for TPU-native performance, optimizing agentic workflows and reasoning loops, authoring reference implementations in PyTorch, JAX, or Pallas, and partnering with TPU hardware, XLA compiler, and cloud AI leadership. The position requires a bachelor's degree or equivalent practical experience, extensive C++ or Python programming experience, software-product launch experience, performance-systems experience, and software-design experience, with compensation of $207,000–$300,000 USD plus a 20% bonus target, equity, and benefits.