Staff ML Compiler Engineer

The Role As a Staff Compiler Engineer on the AI Kernels & Compilers team, you will own the end-to-end compilation stack that takes high-level models and turns them into highly optimized inference artifacts running on GM's autonomous and assisted driving platforms. You'll define the technical vision and build the tooling that makes that path fast, reliable, and effortless for ML engineers across the AV organization to compile their models. You will design and evolve our model export and compilation pipeline-from capturing high-level model graphs, through intermediate representations and compiler transforms, into accelerator-specific inference engines and their integration with our runtime-so that we can simultaneously optimize compilation throughput, model fidelity, and on-vehicle latency. Along the way, you'll build robust tooling to validate numerical correctness, detect and bisect performance regressions, and surface clear, actionable diagnostics back to model authors. What You'll Do - Own and evolve the model compilation toolchain used to deploy large-scale perception, prediction, and planning models to the AV. - Architect new compiler passes and analysis that improve build times, memory footprint, and runtime latency while preserving-or intentionally trading off-fidelity under strict safety and reliability constraints. - Collaborate closely with kernels, runtime, and hardware teams to co-design interfaces, shape accelerator capabilities, and ensure the compiler exposes the right abstractions to unlock peak performance on each platform. - Set standards and best practices for model export, validation, and debugging so that AV teams can iterate quickly with clear, reproducible performance and accuracy characteristics. Your Skills & Abilities (Required Qualifications) - 5+ years of experience in the field of compilers - Experience with ML frameworks (e.g., PyTorch, TensorFlow, JAX) and software stack (e.g., ONNX, MLIR, XLA, TVM, TensorRT, etc) - Expertise in writing production quality Python/C++ code - Expertise in the software development life-cycle - coding, debugging, optimization, testing, integration - BS, or higher degree, in CS/CE/EE, or equivalent What Will Give You A Competitive Edge (Preferred Qualifications) - Experience building and optimizing ONNX-based model export and deployment pipelines - GPU programming (CUDA) and familiarity with ML SW stack (e.g., cuDNN, cuBLAS) - Experience with ML accelerators and hardware architecture - Experience developing and deploying machine learning models