About this role
The Qualcomm Cloud Computing team is seeking a Director of Engineering to lead high-performing teams in the development of hardware and software for Machine Learning solutions across data center, edge, infrastructure, and automotive markets. This role involves managing the full product life cycle from early design to commercial deployment.
Key Responsibilities:
- Lead teams in machine learning software engineering and system design.
- Analyze and optimize training and inference for deep learning models, specifically for Qualcomm AI 100.
- Build software tools and ecosystem around the AI software stack, including abstraction layers for inference accelerators using vLLM, Triton, ExecuTorch, Inductor, and TorchDynamo.
- Optimize workloads for scale-up (multi-SoC) and scale-out (multi-card) systems and improve the deep learning pipeline including graph compiler integration.
- Interface with software development, integration, and test teams to resolve system-level issues.
Qualifications
- Bachelor's, Master's, or PhD in Engineering, Machine Learning/AI, Information Systems, Computer Science, or a related field.
- Minimum experience requirements vary by degree: 8+ years (Bachelor's), 7+ years (Master's), or 6+ years (PhD) in Software Test Engineering or related work.
- 5+ years of experience with software/system test, test plan automation, and CI/CD tools.
- Strong foundation in mathematical modeling, linear algebra, and state-of-the-art ML/AI algorithms.
- Excellent C/C++/Python programming skills with experience in debugging and performance analysis.
- Experience with PyTorch and TensorFlow software stacks.
Preferred Qualifications
- Knowledge of tiling and scheduling machine learning operators.
- Experience with C++ 14, SIMD, multi-threaded high-performance code, ML compilers, and auto-code generation (MLIR).
- Experience running workloads on large-scale heterogeneous clusters.
- Hands-on experience with CUDA and cuDNN.
Skills for this role
Machine LearningDeep LearningC++PythonPyTorchTensorFlowvLLMTritonExecuTorchInductorTorchDynamoCUDAcuDNNMLIRMathematical modelingLinear algebraPerformance analysisSoftware optimization