Software Engineer - Host and Network IO

About the role

Cerebras Systems builds the world's largest AI chip, which delivers training and inference speeds over 10 times faster than GPU-based cloud services. The Host and Network IO team develops the full IO path from a distributed system of server nodes, through the cluster, down to the custom RoCE network stack and the proprietary IOs onto the WSE. You will interface with AI application-level IO teams, cluster architecture teams and FPGA/ASIC teams to optimize bandwidth and latency while minimizing congestion and other issues.

Key responsibilities

• Develop x86 and ARM software to expose next-generation hardware IO capabilities for AI/HPC application teams.
• Govern a generic IO API used by multiple internal users.
• Develop control and configuration subsystems that interact directly with Cerebras hardware.
• Debug network performance in large AI clusters and gather and analyze network statistics and packet traces to find and fix bottlenecks.
• Develop tools and telemetry to increase visibility into the network and IO datapath.
• Optimize CPU and memory utilization using kernel bypass and zero-copy techniques.
• Integrate leading-edge networking technologies and protocols.
• Lead cross-functional technical projects and foster clear communication across teams.

Required profile

• Master's or PhD in Computer Science or Electrical Engineering plus 1 year of industry experience, or 3+ years of industry experience.
• Experience in large software environments.
• Detail-oriented, willing to learn the bigger picture and step outside your comfort zone.

Required skills

• Embedded systems, HW/SW co-design and some driver development.
• Familiarity with network protocols (TCP, RoCE) and network debug tools such as Wireshark, or willingness to learn.
• Some familiarity with network switch environments (e.g., Arista, Juniper) or willingness to learn.