Google's Tensor TPU compiler team is looking to improve the quality and performance of compiler optimizations, correctness, and compilation time for ML workloads, specifically enabling efficient execution of Generative AI models and on-device AI like Gemini Nano on Pixel phones.
Requirements
- 2 years of experience with software development in one or more programming languages (e.g. C++)
- 2 years of experience working with compilers (LLVM, MLIR, etc.).
- 2 years of experience with data structures or algorithms.
- Experience with compiler development in the context of accelerator-based architectures.
- Experience in optimizing ML models for inference.
Responsibilities
- analyzing and improving the compiler quality and performance on optimization decisions, correctness and compilation time.
- Work with Tensor TPU architects to design future accelerators, the HW/SW interface, and co-optimizations of the next generation Tensor TPU architectures.
- Develop parallelization and scheduling algorithms to optimize compute and data movement costs to execute ML workloads on the Tensor TPU.
- Work on efficient mapping of Generative AI (GenAI) models and other key workloads into Tensor TPU instructions through the compiler.
- Collaborate with ML model developers, researchers, and Tensor TPU hardware/software teams to accelerate the transition from research ideas to exceptional user experiences running on the Tensor TPU.
- analysis, optimization, and compilation of Machine Learning (ML) models targeting the Tensor TPU.
- productize the latest ML innovations and research by delivering computing hardware and software.
Other
- Bachelor’s degree or equivalent practical experience.
- Master's degree or PhD in Computer Science or related technical fields.
- The US base salary range for this full-time position is $141,000-$202,000 + bonus + equity + benefits.
- preferred working location from the following: Mountain View, CA, USA; Kirkland, WA, USA
