Engineering Manager, ML Performance

Overview As Engineering Manager, you lead a team of software engineers focused on ML training and serving benchmarks representative of Google production and the ML industry. You drive performance improvements for customer launches and for OSS benchmarks, collaborating with partner teams to seize optimization opportunities. You shape algorithmic innovations leveraging TPU hardware features and model-preserving optimizations, and co-design TPU-friendly models to showcase quality beyond GPU-focused baselines. You guide cross-functional execution across hardware and software domains to scale ML workloads at hyperscale. Compensation / Benefits20% bonus targetequitybenefitsbase salary range: US: $207000 - $300000 ResponsibilitiesLead and mentor a team of engineers focused on ML benchmarks and performance optimizationsDrive near-term SOTA and out-of-the-box performance through compiler/runtime optimizations, tooling, and auto-sharding in collaboration with partner teamsContribute to algorithmic innovations exploiting new TPU hardware features and model-preserving optimizations (e.g., speculative decoding, sparsity, quantization, LoRA)Co-design TPU-friendly models to showcase performance and quality relative to OSS models typically designed on GPUsLead benchmarking efforts to identify performance opportunities and influence hardware and software design decisionsEngage with cross-functional teams across information retrieval, AI, distributed systems, and data centers to deploy large-scale projectsAlign ML performance work with Google’s hardware-accelerated AI strategy and product roadmapsManage project goals, priorities, and budget, reporting progress to leadership Key requirementsBachelor’s degree or equivalent practical experience8 years of software development experience5 years of experience leading ML design and optimizing ML infrastructure3 years of technical leadership experience2 years of people management or team leadership experienceExperience with ML performance analysis, benchmarking, and computer architectureCross-functional collaborationStrategic thinking and roadmap planningPeople leadership and mentoringML performance analysis and benchmarkingComputer architecture knowledgeExperience with ML accelerators (GPUs, TPUs)