Software Engineer, Inference Platform

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 Inference Platform team owns the orchestration layer that runs inference on datacenter clusters, connecting cloud components with machine learning services. The team is often the first to solve unsolved problems, working across Kubernetes operators, service security policies and CI/CD. The role is open to Sunnyvale or Toronto.

Key responsibilities

• Design, develop, test and maintain production software covering testing, continuous development, observability, security, networking, debugging and productionization.
• Help shape the technical direction of the Inference Platform, Kubernetes custom resource definitions, failure domains, service boundaries and system evolution, and own the roadmap for major technical areas.
• Architect active-active systems with rapid failover, graceful degradation and clear SLOs, improving latency, throughput, capacity efficiency and resilience.
• Write and review production code in critical parts of the platform, make high-consequence architectural decisions and lead design and code reviews.
• Lead on the hardest production issues and bottlenecks, driving observability, incident response, capacity planning and post-incident improvement.
• Partner with ML, Product, Infrastructure and Cloud teams to turn product and business needs into scalable system designs.

Required profile

• 3+ years of software engineering experience, building and operating large-scale distributed systems or cloud infrastructure.
• Experience with distributed systems, ideally with Kubernetes.
• Experience building highly available, latency-sensitive systems at scale.
• Experience with security (certificates, TLS, mTLS).
• Experience optimizing latency, throughput and efficiency in high-QPS systems; TTFT and tail-latency reduction is a strong plus.

Required skills

• Backend or systems languages such as Go or C++.
• Preferred: ML inference infrastructure, model serving systems or GPU-accelerated workloads.