ByteDance
Posted 1y ago

硬件加速推理引擎运行时开发工程师-AI工具链

ByteDance
Beijing, Beijing, China
OnsiteFull Time
Responsibilities
  • designing runtime
  • implementing components
  • optimizing graphs
Requirements
  • Strong C++ skills
  • Solid CS/EE fundamentals
  • Familiarity with computer architecture
  • Data structures and algorithms
  • Experience with heterogeneous runtimes or driver development preferred
  • CUDA/ROCm and Python experience a plus
Technical tools mentioned
C++CUDAAMD ROCmPython

Job description

1、设计并实现推理引擎的核心运行时组件,包括模型加载、图优化、算子调度、内存管理等;
2、面向深度学习AI芯片的Runtime/UMD软件栈的设计、开发与日常维护。

1、计算机科学/电子工程等相关专业,精通C++编程语言;
2、计算机专业基础扎实,熟悉计算机体系结构、常用数据结构与算法;
3、具有异构Runtime、驱动开发经验者优先;
4、熟悉CUDA Runtime,AMD ROCm/CLR等软件,对CUDA兼容Runtime/UMD有过开发经验者优先;
5、对新技术充满热情,能够快速在新的领域进行开拓并输出结;
6、团队沟通协作能力强。

加分项
1、熟悉GPU体系结构NPU和常见异构编程模型;
2、了解IC实现细节;
3、有AI相关基础知识;
4、熟悉Python语言。

About ByteDance

Developing AI-driven content platforms and mobile applications.