Network Engineer

Cerebras Systems•Sunnyvale, CA

About The Position

Cerebras Systems is seeking a Network Architect to join their Cluster Engineering Team. This role will focus on shaping the front-end datacenter and interconnect fabric for current and future AI clusters. The architect will collaborate with hardware vendors, internal networking teams, and industry peers to design resilient, reliable, and high-throughput network architectures for large-scale AI workloads. This is a technical role requiring full-stack knowledge, from host-side networking and NIC behavior to cluster-level coordination, with expertise in various hardware components and their software layers. The position involves leading proof-of-concept work for new network designs, acting as the central technical authority for network reliability, and driving innovation in network architecture.

Requirements

Ph.D. in Computer Science or Electrical Engineering with 5+ years of industry experience, or Master's in CS/EE with 10+ years of industry experience.
Solid experience in designing large-scale networks in datacenter and cloud environments.
Extensive hands-on experience debugging networking issues in large distributed systems with multiple platforms and protocols.
Demonstrated track record leading multi-phase, multi-team technical projects to completion.
Deep expertise across networking platforms: Juniper, Arista, Cisco, and open-box / disaggregated NOS architectures (SONiC).
Strong working knowledge of networking protocols and fabric technologies: VXLAN, EVPN, RoCEv2, BGP, DCQCN, PFC, ECN, and streaming telemetry.
Proficiency in Python (and/or Go) for building network automation, validation, and tooling.
Comfort with config generation frameworks (Ansible, Jinja2), gNMI, and CI/CD pipelines for network infrastructure.
Hands-on experience with streaming telemetry pipelines, time-series databases (Prometheus, InfluxDB), visualization (Grafana), log aggregation, and modern incident-management.
Ability to define SLIs/SLOs and instrument the network for proactive reliability.
Familiarity with network visibility, management, and packet-capture/analysis tools.

Nice To Haves

Prior experience at hyperscalers or cloud service providers.
Experience with AI/ML or HPC cluster networking, including lossless Ethernet design, rail-optimized topologies, and collective-communication traffic patterns.
Track record of contributions to open-source networking projects, standards bodies, or industry conferences.

Responsibilities

Design and architect front-end network fabrics for AI/ML and HPC clusters, optimizing for high resource utilization, low latency, and high-throughput communication.
Build proof-of-concept implementations of new network designs and features, and drive them from prototype through production rollout.
Identify and resolve performance and efficiency bottlenecks across the host-NIC-fabric.
Automate the deployment, configuration, and validation of network infrastructure using Python, including topology provisioning, fabric bring-up, config generation, and regression.
Stand up and operate SRE-grade telemetry and observability for the cluster network: streaming telemetry (gNMI, OpenConfig, sFlow/IPFIX), metrics pipelines, alerting, and incident workflows.
Define the SLIs/SLOs that govern network reliability and drive blameless post-incident analysis.
Lead network debugging in large distributed-systems environments spanning multiple platforms and protocols, including deep dives into RoCEv2, PFC/DCQCN, ECMP hashing, congestion behavior, and packet-level forensics.
Lead cross-functional, multi-phase technical projects spanning hardware, firmware, host networking, and cluster software.
Collaborate with vendors and industry partners to shape network hardware and feature roadmaps.
Represent the company in industry forums, standards bodies, and technical communities.
Serve as the central point of contact for network reliability issues across the cluster.