tvm jobs in San Diego

Compiler Software Engineer responsible for designing and enhancing compiler toolchains and development tools for Qualcomm's mobile and compute products across gaming, image processing, AR/VR, and machine learning domains. The role involves working with processor architectures (DSP, CPU, NSP), optimizing code compilation, building simulators and debuggers, and applying mathematical techniques for performance and power efficiency.…See more Will assist in developing industry-standard tools using frameworks like LLVM, GCC, MLIR, and TVM, and work on intermediate representation and program optimization. Works under supervision on compiler infrastructure and programming tools to support software developers.

Design and develop the end-to-end compilation pipeline for Qualcomm AI Hub Workbench, enabling PyTorch and ONNX models to compile into deployable artifacts targeting LiteRT, ONNXRuntime, and QAIRT across Snapdragon SoCs. Own graph optimization, backend dispatch across CPU/GPU/NPU, model catalog validation, and automated CI/CD pipelines for on-device ML deployment.…See more Requires 4+ years of software/hardware/systems engineering experience and proficiency in Python and C++, with preferred expertise in ML compiler infrastructure, MLIR/ONNX/TVM, torch.export, and on-device frameworks.

Staff Machine Learning Engineer role focused on designing and maintaining Qualcomm AI Hub's end-to-end ML compilation pipeline, from PyTorch and ONNX model ingestion through graph optimization to deployable artifacts on Snapdragon SoCs. Responsibilities span compiler infrastructure (graph transformation, op validation, backend dispatch across CPU/GPU/NPU), model catalog automation and CI/CD, and developer tooling for profiling and debugging.…See more Requires 4+ years of software/systems engineering experience (or 3+ with a master's, or 2+ with a PhD), with preferred expertise in ML compiler concepts, ONNX/PyTorch export workflows, and on-device deployment frameworks.