Senior AI Systems Engineer - Kernel & GPU Networking

Cisco; Milpitas (CA); On-site; USD 195,000 - 218,000; Full time

Job summary

Cisco is seeking a Senior Software Engineer to join the CVIS team in Milpitas, CA, responsible for debugging, profiling and optimizing the AI cluster networking and GPU performance stack from boot-time through the Linux kernel, RDMA, NCCL and application-level communications.

You will investigate kernel-level failures, boot-time anomalies, PCIe issues, and driver behavior, diagnose and tune RDMA/RoCEv2, NCCL, SmartNIC/DPU, and NIC queues, and read traces, logs and observability data to identify

Qualifications

• 4+ years of experience working with Linux kernel, networking, GPU, and distributed-systems debugging.
• Proficient in C/C++/Python and driver development (user-mode/kernel-mode).
• Experience with Linux networking stack, RDMA/RoCEv2, NCCL, PCIe and performance profiling.

Responsibilities

• Investigate kernel-level failures, boot-time anomalies, PCIe link issues, driver behavior, and network latency.
• Diagnose and tune RDMA/RoCEv2, NCCL, SmartNIC/DPU, NIC queues, congestion, throughput, and collective communications.
• Read and correlate unified hardware traces across host CPUs, PCIe switches, SmartNICs, and GPU compute streams.
• Capture and analyze Wireshark packets, system logs, NVIDIA support logs, traces, profiles, and observability data.
• Partner with hardware and software teams to eliminate end-to-end bottlenecks.

Skills

Linux kernel

Networking

GPU

Distributed systems debugging

C/C++/Python

RDMA

NCCL

PCIe

Performance profiling

Education

Bachelor's degree in Computer Science or related field

Tools

Nsight profiling tools

Job description

Cisco is seeking a Senior Software Engineer to join the CVIS team in Milpitas, CA, responsible for debugging, profiling and optimizing the AI cluster networking and GPU performance stack from boot-time through the Linux kernel, RDMA, NCCL and application-level communications.

You will investigate kernel-level failures, boot-time anomalies, PCIe issues, and driver behavior, diagnose and tune RDMA/RoCEv2, NCCL, SmartNIC/DPU, and NIC queues, and read traces, logs and observability data to identify