HiringCafe
Witmem
Posted 1mo ago

算子开发工程师（异构计算方向）

Witmem
witintech.comView all open roles
Beijing or Hangzhou or Shanghai
OnsiteFull Time
Responsibilities
  • developing operators
  • tuning performance
  • integrating framework
Requirements
  • Bachelor's in computer/electrical/automation
  • 3+ years C++ experience
  • Multi-threading
  • GPU programming (CUDA/OpenCL)
  • Deep learning operator knowledge
  • NPU operator experience, and performance tuning ability
Technical tools mentioned
C++CUDAOpenCL

Job description

面向自研异构计算平台，负责三类硬件后端的算子开发：
1. CPU 算子 ：推理/量化算子的高性能实现
2. GPU 算子 ：推理/量化算子的 GPU 版本，含显存管理与数据搬运
3. 定制 NPU 算子 ：基于定制编程模型，实现模型推理算子，适配专用指令集与片上存储
4.配合团队完成算子接入调度框架、性能调优与基准测试。

1.本科及以上，计算机/电子/自动化等相关专业，3 年以上 C++ 开发经验
2.熟练掌握现代 C++ 与多线程编程
3.具备 GPU 编程实战经验（CUDA / OpenCL 等均可）
4.熟悉常见深度学习算子原理与优化思路
5.有至少一种定制加速器 / NPU 的算子开发经验
6.具备性能调优能力与工程化思维

About Witmem

Develops computing-in-memory AI chips for edge devices.

Year founded
2017
Employees
300
Industries
SemiconductorsArtificial Intelligence
Activities
Computing-in-memory chipsAI processor designEmbedded system integrationSemiconductor R&DNeural network hardware
Organization type
Private
Latest investment
Raised $200.00M Series B (2023) — led by 国投创业
Headquarters
CN

Similar jobs

Operator Development Engineer roles near Beijing, Beijing
4mo
Save
Mark Applied
Hide
算子开发工程师
Beijing or Shenzhen
OnsiteFull Time
Qingcheng.AI
Qingcheng.AI: Develops high-performance AI inference engines and deployment software.
4+ YOEMaster's degree in CS or related, 4+ years experience, proficiency in Python/C++/C, CUDA or AscendC, GPU operator optimization and inference-engine co-optimization experience.
Python, C++, C, CUDA, AscendC, PyTorch
Job PostingView all
2w
Save
Mark Applied
Hide
AI算子开发工程师
Beijing or Xi'an or Shanghai
OnsiteFull Time
Xiaomi
XiaomiHKEX: 1810: Designs and manufactures smartphones, consumer electronics, and smart devices.
Bachelor's degree or higher in computer science, electronic engineering, mathematics, or related fields; proficiency in CUDA, C++, or Python; CUDA-like or NPU SDK development experience; parallel computing knowledge preferred.
CUDA, C++, Python, OpenCL, HIP, NPU SDK, Nsight, VTune
Job PostingView all