Principal Engineer, Inference Cloud

Responsibilities

• Identify and prioritize critical technical problems for the Inference Cloud Platform, making explicit tradeoff decisions regarding platform support and evolution.
• Set the long-term technical direction for the platform, including multi-region topology, failure domains, and service boundaries.
• Architect highly available, active-active systems featuring rapid failover and graceful degradation mechanisms like circuit breaking and load shedding.
• Drive continuous improvements in latency, throughput, capacity efficiency, and resilience under unpredictable, bursty AI workloads.
• Contribute production code to critical paths and lead design and code reviews to ensure high engineering standards.
• Lead incident response, observability, and capacity planning to maintain high operational rigor and reliability.
• Mentor engineers and influence technical strategy across adjacent teams regarding API design, deployment strategy, and shared infrastructure.

Requirements

• 10+ years of software engineering experience, with a focus on building and operating large-scale distributed systems or cloud infrastructure.
• Deep expertise in distributed systems architecture, including networking, compute orchestration, and multi-region production services.
• Proven track record of making sound architectural decisions for highly available, latency-sensitive systems at scale.
• Experience optimizing performance in high-QPS systems, specifically regarding latency and throughput.
• Strong proficiency in backend or systems programming languages such as Go, C++, or Python.
• Experience designing and implementing observability and reliability practices (metrics, logging, tracing, and SLI/SLO-driven operations).
• Ability to influence senior technical leadership and cross-functional partners through technical credibility and judgment.

Preferred Qualifications

• Experience with ML inference infrastructure, model serving systems, or GPU-accelerated workloads.
• Specific experience with TTFT (Time To First Token) and tail-latency reduction.

About the Company

Cerebras Systems builds the world's largest AI chip, providing the compute power of dozens of GPUs on a single wafer-scale architecture. Our technology delivers industry-leading training and inference speeds, empowering users to run large-scale ML applications effortlessly. Cerebras Inference offers the fastest Generative AI inference solution in the world, transforming the user experience for top model labs and global enterprises.