AI计算库开发工程师(北京/深圳)
Core
Develop and optimize high-performance operators and design efficient data/computation flow schemes for self-developed AI chips.
Role type
Senior IC AI compute library engineer (hardware/software stack)
Builds
High-performance operator implementations and optimized data/computation flows for self-developed AI chips
Domain
AI hardware acceleration / High-performance computing
Deliverable
production ML models
Required skills
C++, Python, Linux development, heterogeneous computing, parallel programming, vectorization, memory access optimization, PyTorch, TensorFlow, performance profiling and debugging
Preferred skills
Compiler technology, runtime optimization, end-to-end model optimization
Technologies
PyTorch, TensorFlow, C++, Python, Linux
Responsibilities
Develop and optimize high-performance operators based on self-developed AI chip software stacks; Build maintainable and scalable operator implementations leveraging chip compute units and memory hierarchy; Participate in building test frameworks and performance analysis systems; Collaborate with compiler, runtime, and business teams to drive end-to-end performance optimization and deployment of models on self-developed chips.
