You will lead multiple teams responsible for running state-of-the-art models (LLaMA, DeepSeek, diffusion, MoE and future generations of AI), building high-performance kernels, integrating industry frameworks, and delivering a developer-friendly stack to customers, which showcases and materializes the full potential of our hardware.
Responsibilities:
Own the vision, roadmap, and execution of the full AI software stack (frameworks, compilers, runtime, kernels, tools).
Lead teams responsible for running and optimizing AI workloads
Deliver first-class support for PyTorch, Triton, IREE, CUDA-compatible flows, and emerging AI frameworks.
Oversee development of high-performance kernels (GEMM, attention, memory ops) and compiler optimizations.
Drive HW/SW co-design with silicon, architecture, and runtime teams to maximize performance and efficiency.
Ensure end-to-end product delivery, including SDKs, toolchains, developer experience, and documentation.
Own benchmarking, performance targets, and competitive analysis
Build, mentor, and scale a multi-disciplinary AI software organization.
Support bring-up, validation, and tuning of new hardware accelerator generations (pre-silicon and post-silicon).
Bachelors or Masters degree and/or equivalent experience in computer science or a related field.
5+ years in AI/ML systems, GPU/accelerator software, or ML frameworks.
10+ years managing multiple teams or leading large AI software efforts.
Deep expertise in PyTorch internals, Triton, IREE, XLA, ONNX, or similar AI compiler stacks.
Strong background in kernels (GEMM, attention), performance engineering, and accelerator bring-up.
Proven ability to deliver production-grade AI software stacks on new hardware.
Knowledge of dataflow architectures, or heterogeneous compute is an advantage.
Experience with customer deployments, SDK delivery, and developer ecosystems.
Track record of founding or scaling high-performance engineering organizations.
