System Software Engineer Scalable Solution

You will design and implement components of an NPU communication software stack, including a collective communication library and driver. You will contribute to API and performance design across software and hardware layers, analyze bottlenecks, and influence future NPU and interconnect topology.

Responsibilities

• Design and implement components of an NPU communication software stack, including collective communication libraries and drivers
• Contribute to technical design, API definition, and performance optimization across software and hardware layers
• Analyze performance bottlenecks with hardware and software teams
• Influence future NPU and interconnect topology

Requirements

• At least 5 years of systems software development experience or a related Ph.D.
• C/C++ low-level systems programming
• Operating system internals and networking knowledge
• Experience delivering complex, high-performance, reliable software
• Hardware accelerator and performance characteristic knowledge
• Collective communication algorithm knowledge, including All-Reduce, All-Gather, and Reduce-Scatter
• High-performance communication library, parallel runtime, or high-performance networking experience
• RDMA/RoCE, NVLink, or CXL knowledge
• Interconnect topology and Network-on-Chip architecture knowledge