1 day ago
Shanghai, China or Chengdu, ChinaSenior
Responsibilities
- Develop features and fix bugs in Qualcomm Neural Network SDK and AI Engine Direct SDK.
- Develop neural network operator kernels using Hexagon DSP or SIMD accelerators.
- Optimize performance and tune model accuracy using Qualcomm AI Toolchains.
- Design and implement algorithms for standard and user-defined neural network operators.
- Apply quantization-aware training and post-training quantization.
- Collaborate with regional and cross-functional teams on feature design and implementation.
- Support testing teams in improving AI Toolchain quality.
- Assist customers with efficient neural network deployment.
- Optimize deep learning inference for Qualcomm CPU, GPU, and DSP hardware platforms.
Requirements
- 5–6 years of software development experience and the ability to independently complete projects on edge platforms.
- Bachelor's degree with 4+ years, master's degree with 3+ years, or PhD with 2+ years of relevant software engineering experience.
- 2+ years of work experience with programming languages such as C, C++, Java, or Python.
- Experience with SDK development for varied system platforms and familiarity with ARM-based hardware and software.
- Preferred experience designing and implementing deep learning networks with modern frameworks.
- Preferred familiarity with ONNX QDQ, PyTorch, TensorFlow, PPQ, and TensorRT quantization frameworks.
- Preferred experience with model accuracy analysis, software design, debugging, documentation, and presentation.
- Preferred experience with QNN, TfLite, NCNN, TNN, or MACE inference frameworks.
- Preferred experience with OpenCL, NEON, OpenMP, or CUDA parallel programming.
- Preferred DSP software development and algorithm implementation experience, including Hexagon or HVX familiarity.
- Experience with GenAI model quantization and inference acceleration, or deployment of ADAS/VLA, GenAI/LLM/LVM, ASR, or NLP models, is a plus.
- Strong problem-solving, logical thinking, communication, and cross-functional collaboration skills are required.
About Qualcomm
Inspired by 250 years of American ingenuity, we build technologies that expand what’s possible. From wireless to AI at scale and 6G, our innovations power progress worldwide.