Distributed Software Engineer
Cerebras builds wafer-scale AI computing systems and a cloud inference platform for training, fine-tuning, and serving AI models.
About Cerebras Systems, Inc.
Cerebras Systems is an AI-infrastructure company founded in 2015. It sells rack-scale wafer-scale computing systems and provides cloud-based, API-accessible AI inference alongside on-premises deployments.
Skills
Candidate Availability
Required and preferred rules are kept separate and reflect the wording in the original posting.
About the Role
You will build and operate software that turns large fleets of Cerebras systems, servers, and switches into reliable, observable clusters. You will automate bare-metal infrastructure, create Kubernetes operators and control-plane services, improve fleet reliability, and expose cluster capabilities through APIs, CLIs, and an MCP gateway.
Requirements
- 5+ years building and operating production distributed systems or infrastructure software
- Production-quality Go and Python skills
- Experience writing or debugging Kubernetes controllers and operators
- Knowledge of CRDs, reconciliation semantics, informer caches, admission webhooks, and RBAC
- Debugging skills across distributed systems, Linux, and networking
- Experience with Prometheus, Grafana, PromQL, exporter design, and alerting
- Active use of coding agents and rigor in verifying their output
Responsibilities
- Build declarative CRD-driven automation for bare-metal networking, operating systems, and application software across clusters
- Deliver push-button cluster installation, upgrades, and security patching with canaries and downtime budgets
- Develop Kubernetes operators for scheduling large inference workloads
- Build gRPC control-plane services, authorization, admission webhooks, and quota policies
- Create metrics and log pipelines, exporters, SLOs, and alerting for systems, servers, and network fabric
- Implement failure detection, highly available control planes, and automated recovery
- Develop CLIs, APIs, and an MCP gateway for users, operators, and AI agents