About The Role

Join Cerebras as a Performance & Reliability Engineer within our innovative Co-Design and Next Generation Team. Our groundbreaking CS-3 system has set new benchmarks in high-performance ML training and inference solutions. It leverages a dinner-plate sized chip with 44GB of on-chip memory to surpass traditional hardware capabilities. This role focuses on characterizing and optimizing the performance and reliability of state-of-the-art AI models running on Cerebras' breakthrough hardware.

Responsibilities

Characterize and enhance the performance and reliability of advanced ML hardware/software systems, with emphasis on reducing power and thermal fluctuations.
Analyze ML workloads, software kernels, and hardware architecture for power and performance impacts, and synthesize high-level insights across these layers.
Develop creative software solutions to improve reliability and performance, collaborating cross-functionally to deploy these solutions in production.
Influence the design of Cerebras' next-generation AI architecture and software stack through rigorous workload analysis and computational efficiency optimization.
Partner with ML engineers, researchers, and reliability specialists to understand model behavior and drive system-level improvements from a software perspective.
Collaborate with teams in architecture, silicon, and research to advance our computational platforms and influence future system designs.

Skills & Qualifications

BS, MS, or PhD in Computer Science, Electrical Engineering, or a related field.
3+ years of relevant experience in performance engineering, reliability, computer architecture, and/or software design.
Proficiency in Python or other scripting languages.
Experience with C/C++ and assembly programming.
Demonstrated expertise with system-level performance and reliability optimization.
Strong verbal and written communication skills.
Nice to have: Hands-on experience with ML models, ML frameworks, and collective communication.
Nice to have: Understanding of thermal management principles and power delivery for advanced semiconductors.

Performance & Reliability Engineer

Description

Stack