Senior ML Performance Engineer for GenAI & LLMs

DeepMind Technologies Limited; Mountain View (CA); On-site; USD 174,000 - 252,000; Full time

Job summary

Google DeepMind in Mountain View is seeking a Software Engineer to advance AI agents, focusing on performance and scalability of Gemini inference across TPU platforms. You will build systems for agent testing, visualization, and production-ready pipelines, collaborating with ML researchers and hardware teams to push the boundaries of AI capabilities.

The role emphasizes reducing latency and cost per token while supporting large-scale experiments, with a strong software engineering foundation and

Qualifications

• Bachelor’s degree in CS, software engineering, mathematics, or related STEM field or equivalent practical experience.
• 5 years of experience in software development, systems programming, or applied ML engineering.
• 3 years of experience with performance optimization, data analysis, or debugging.
• 3 years of experience with Generative AI/LLMs and ML frameworks.
• 1 year designing and launching production software or foundation AI systems.

Responsibilities

• Drive down the cost per token of Gemini inference by optimizing throughput, accelerator utilization, batching, scheduling, and KV cache efficiency.
• Be among the first to bring Gemini to new TPU generations by characterizing hardware at launch, tuning kernels, sharding, and serving configurations.
• Implement model compression techniques such as quantization and sparsity to reduce the cost-per-token for production Gemini models.

Skills

Software development

Performance optimization

Data analysis/visualization

Generative AI/LLMs

Production AI systems

Education

Bachelor's degree in CS/Math/related STEM

Master's or PhD in CS/AI/Robotics

Tools

PyTorch

JAX

Triton

vLLM

Gemini API

Job description

Google DeepMind in Mountain View is seeking a Software Engineer to advance AI agents, focusing on performance and scalability of Gemini inference across TPU platforms. You will build systems for agent testing, visualization, and production-ready pipelines, collaborating with ML researchers and hardware teams to push the boundaries of AI capabilities.

The role emphasizes reducing latency and cost per token while supporting large-scale experiments, with a strong software engineering foundation and