Principal NPU Architect - Hybrid

JOB AD:

• Develop and Optimize the hardware architecture for Ascend Core, focusing on performance, power efficiency and scalability.
• Design custom ISA and hardware components that effectively support AI algorithms and functions; balancing performance, power consumption and areas trade-offs Participate in design review and providing technical guidance to team members.
• Stay up-to-date with the latest trends and advancements in AI algorithms, AI hardware and semiconductor technologies.

Collaboration & Project Management

• Collaborate with software teams to ensure seamless integration of Ascend Core Chipset with AI frameworks, compliers and running system.
• Work closely with cross function team to ensure the successful development and deployment of Ascend Core Products.

Skills/Qualifications

Essential

• Rich experience in CPU or GPU architecture and micro-architecture concepts Rich experience in ISA development, caching, or memory subsystem design Familiar with ML framework such as PyTorch, Tensorflow Proficient in VHDL/Verilog Proficient in C/C++ and scripting language such as Perl, Python and Ruby

Desirable

Logical Design background

• Accelerating DNN network, particular on transformer network and GNN Hands on Experience with architecture modelling using higher level language such as C/C++ Knowledge of complier backend, such as code generation, performance analysis and code Optimization