Senior Deep Learning Compiler Engineer - PyTorch
Core
Design, implement, and optimize the Thunder source-to-source compiler to accelerate PyTorch models on NVIDIA GPUs.
Role type
Senior Deep Learning Compiler Engineer
Builds
Thunder compiler stack for PyTorch 2.0 (TorchDynamo, TorchInductor)
Domain
AI compilers, deep learning frameworks, high-performance computing
Deliverable
production ML models
Required skills
Python, deep learning frameworks (PyTorch, JAX), compiler concepts (ASTs, IR, SSA, code generation), performance analysis, CUDA programming
Preferred skills
Contributions to compiler projects (TVM, MLIR, IREE), PyTorch internals expertise, JAX functional transformations, distributed systems, open-source community leadership
Technologies
PyTorch, TorchDynamo, TorchInductor, Thunder, nvFuser, TVM, XLA, CUDA, MLIR, IREE
Responsibilities
Lead design and implementation of core compiler technologies, optimize deep learning workloads across thousands of GPUs, collaborate with compiler and systems teams, translate research into practical solutions
Seniority
Senior, hands-on IC