Solution Architect - GPU/TPU Kernel Optimization

As a Solution Architect specializing in GPU/TPU Kernel Optimization, your role will involve designing and optimizing high-performance kernels for cutting-edge Machine Learning operations. You will be responsible for redefining performance boundaries across massive training runs and high-speed inference workloads while also shaping the developer infrastructure that powers next-generation AI systems. Key Responsibilities: - Design and optimize high-performance kernels targeting Tensor Processing Unit (TPU) and Graphics Processing Unit (GPU) architectures for critical Machine Learning (ML) operations using languages like Pallas, Mosaic, and Triton. - Architect infrastructure such as benchmarking suites, autotuning frameworks, performance analysis tools, regression testing, and documentation. - Transform the developer community's interaction with custom kernels in key Open-Source Software (OSS) libraries. - Stay updated on advancements in hardware architectures, compiler technologies, and AI models to identify new opportunities for performance optimization through custom kernels. - Collaborate with ML researchers, framework developers (JAX, PyTorch), and compiler engineers (XLA) to enhance adoption and address bottlenecks with appropriate solutions. Qualifications Required: - 12-18 years of experience in software development. - Expertise in optimizing TPU/GPU code using low-level kernel languages like Pallas, CUDA, or Triton. - Knowledge of ML Frameworks (JAX/PyTorch) and common operations like attention and Mixture of Experts (MoEs). - Understanding of modern accelerators and compiler principles, including toolchains such as MLIR and OpenXLA. - Showcase of building developer infrastructure, including Open-Source Software (OSS) libraries and flexible high-performance APIs. - Excellent investigative and problem-solving capabilities with strong communication skills across cross-functional teams. If there are any additional details about the company in the job description, please provide them for a more comprehensive job description. As a Solution Architect specializing in GPU/TPU Kernel Optimization, your role will involve designing and optimizing high-performance kernels for cutting-edge Machine Learning operations. You will be responsible for redefining performance boundaries across massive training runs and high-speed inference workloads while also shaping the developer infrastructure that powers next-generation AI systems. Key Responsibilities: - Design and optimize high-performance kernels targeting Tensor Processing Unit (TPU) and Graphics Processing Unit (GPU) architectures for critical Machine Learning (ML) operations using languages like Pallas, Mosaic, and Triton. - Architect infrastructure such as benchmarking suites, autotuning frameworks, performance analysis tools, regression testing, and documentation. - Transform the developer community's interaction with custom kernels in key Open-Source Software (OSS) libraries. - Stay updated on advancements in hardware architectures, compiler technologies, and AI models to identify new opportunities for performance optimization through custom kernels. - Collaborate with ML researchers, framework developers (JAX, PyTorch), and compiler engineers (XLA) to enhance adoption and address bottlenecks with appropriate solutions. Qualifications Required: - 12-18 years of experience in software development. - Expertise in optimizing TPU/GPU code using low-level kernel languages like Pallas, CUDA, or Triton. - Knowledge of ML Frameworks (JAX/PyTorch) and common operations like attention and Mixture of Experts (MoEs). - Understanding of modern accelerators and compiler principles, including toolchains such as MLIR and OpenXLA. - Showcase of building developer infrastructure, including Open-Source Software (OSS) libraries and flexible high-performance APIs. - Excellent investigative and problem-solving capabilities with strong communication skills across cross-functional teams. If there are any additional details about the company in the job description, please provide them for a more comprehensive job description.