- Mountain View, California, United States
- On-site
- Full-time · Senior
- $174K–$252K per year
- 5+ years
Software Engineer focused on performance optimization of generative AI models and TPU platforms at DeepMind
Compensation
US: $174000 - $252000 (USD) + 15% bonus target + equity + benefits
Responsibilities
- Drive down the cost per token of Gemini inference by optimizing throughput, accelerator utilization, batching, scheduling, and KV cache efficiency
- 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 cost per token
- Evaluate new architectures, optimizers, and training methods to achieve target quality with reduced compute requirements
Requirements
- Bachelor's degree in Computer Science, Software Engineering, Mathematics, or related STEM field
- Master's or PhD in Computer Science, Artificial Intelligence, Robotics, or related technical field
- Nice to have: technical leadership
- Nice to have: architectural direction
- Nice to have: guiding project teams