Staff Software Engineer, Cloud ML Compute Services

Google Cloud is seeking a Staff Software Engineer to join their Cloud ML Compute Services team, focusing on building and supporting Google Cloud Platform's Cloud TPU and GPU services. This role is critical in developing next-generation technologies that impact billions of users' interactions with information and each other. The position requires expertise in machine learning infrastructure, working with frameworks like PyTorch and JAX, and optimizing performance for ML workloads.

The role involves working across the full technology stack, from high-level Python to low-level C++, to improve LLM training and inference performance on TPU. You'll be responsible for implementing new features, publishing high-performance kernels, and collaborating with various teams to enhance PyTorch capabilities and enable new workloads.

As a Staff Software Engineer, you'll be part of Google Cloud's mission to accelerate digital transformation across industries. The team provides enterprise-grade solutions leveraging Google's cutting-edge technology, serving customers in more than 200 countries. You'll work with state-of-the-art ML supercomputers and have the opportunity to impact how organizations access and utilize machine learning capabilities at scale.

The position offers competitive compensation, including a base salary range of $189,000-$284,000, plus bonus, equity, and comprehensive benefits. You'll be working with industry-leading professionals, having the chance to shape the future of machine learning infrastructure while solving complex technical challenges that affect global users.

This role is perfect for someone who is passionate about machine learning, has strong technical leadership experience, and wants to work at the forefront of AI infrastructure development. You'll have the opportunity to influence technical direction, mentor team members, and contribute to Google Cloud's mission of enabling organizations to leverage cutting-edge ML capabilities efficiently and effectively.