Performance Engineer — Hardware Accelerators (TPU/GPUs)

Socket.dev; Mountain View (CA); On-site; USD 147,000 - 210,000; Full time

Benefits offered by this job

Bonus target 15%

Equity

Benefits

Job summary

Google DeepMind is seeking a Software Engineer to advance Gemini model training and serving on hardware accelerators (TPUs/GPUs). You will work across the JAX stack, implement low-level kernels, and optimize performance across distributed systems.

You will profile workloads, design benchmarking pipelines, and contribute to compiler passes and accelerator roadmaps, collaborating with ML and neuroscience teams to push the boundaries of AI while focusing on safety and scalable infrastructure.

Qualifications

• Bachelor’s degree or equivalent practical experience.
• 2 years of experience with software development in C++ or Python.
• Experience with profiling tools (e.g., gProf, Valgrind, or VTune) for performance optimization.
• Experience in software development for hardware accelerators (e.g., CPUs, GPUs, or TPUs), including memory-hierarchy or instruction-level tuning.

Responsibilities

• Improve the training and serving efficiency of Gemini models on hardware accelerators (TPUs and GPUs), spanning model configurations, execution runtimes, and dedicated compiler passes.
• Profile large-scale distributed workloads to diagnose compute, memory, and communication bottlenecks, identifying high-impact optimization opportunities.
• Design and implement high-performance, low-level custom kernels for critical model operators to unlock peak hardware utilization.
• Build and maintain automated benchmarking pipelines and diagnostic tooling to track, reproduce, and guard against performance regressions.
• Influence next-generation accelerator architectures and compiler roadmaps by feeding back empirical workload profiles and model requirements.

Skills

Python or C++

Profiling tools

Hardware accelerators

Machine learning optimization

Education

Bachelor’s degree or equivalent practical experience

Master’s degree or PhD

Tools

gProf

Valgrind

VTune

Job description

Google DeepMind is seeking a Software Engineer to advance Gemini model training and serving on hardware accelerators (TPUs/GPUs). You will work across the JAX stack, implement low-level kernels, and optimize performance across distributed systems.

You will profile workloads, design benchmarking pipelines, and contribute to compiler passes and accelerator roadmaps, collaborating with ML and neuroscience teams to push the boundaries of AI while focusing on safety and scalable infrastructure.