Prognostics & Health Monitoring Engineer

Cerebras Systems•Sunnyvale, CA

About The Position

Quality, reliability, and uptime are foundational to scaling Cerebras systems. We are seeking an engineer to define and build our prognostics and health monitoring (PHM) capability—developing frameworks to monitor, assess, and predict hardware health across our fleet. In this role, you will transform telemetry and operational data into actionable insights and automated responses, enabling early detection of degradation, accurate failure prediction, and proactive actions to keep systems highly available, performant, and resilient. This is a highly cross-functional role spanning reliability engineering, data science, and system software, with broad influence across hardware, software, and fleet operations.

Requirements

Bachelor’s or Master’s in Engineering, Computer Science, Data Science, or related field
8+ years in reliability engineering, data science, fleet analytics, or similar
Strong Python and SQL for large-scale data analysis and modeling
Experience building and deploying predictive models in production
Expertise in applied statistics and probabilistic modeling (e.g., survival analysis, hazard models, Bayesian methods)
Experience with large-scale telemetry or distributed system datasets
Proven ability to define ambiguous problems and deliver scalable solutions

Nice To Haves

Experience with HPC systems, AI infrastructure, or datacenter environments
Background in PHM, predictive maintenance, or reliability analytics at scale
Familiarity with RUL estimation and degradation modeling
Understanding of observability systems, telemetry pipelines, and real-time monitoring
Background in hardware reliability and failure modes in complex systems

Responsibilities

Define the vision, architecture, and roadmap for PHM across deployed systems
Design and scale frameworks for health assessment, anomaly detection, and predictive failure modeling
Develop and productionize probabilistic models for failure risk, degradation, and remaining useful life
Analyze large-scale telemetry, logs, and service data to identify systemic drivers of failures and disruptions
Establish health metrics, scoring systems, and fleet-level observability to communicate system risk
Partner with system software to integrate monitoring, alerting, and automated mitigation into production
Drive closed-loop systems (detection → diagnosis → action → validation)
Influence hardware design, qualification, and operations through data-driven insights