CoDesign & NextGen Performance Engineer
Cerebras Systems2 months ago
Toronto, Canada or Sunnyvale, CA, USAMid Level
H1B sponsor
Responsibilities
- Bring up and optimize performance on new generations of the Cerebras Wafer Scale Engine.
- Build kernel-level and end-to-end performance models for state-of-the-art and customer machine learning models.
- Optimize and debug kernel microcode and compiler algorithms to improve inference speed, throughput, and compute utilization.
- Debug and analyze runtime performance on systems and clusters.
- Develop tools and infrastructure to visualize performance data from the Wafer Scale Engine and compute cluster.
Requirements
- Bachelor’s, master’s, or PhD in Electrical Engineering or Computer Science.
- Strong background in computer architecture.
- Understanding of low-level deep learning and LLM mathematics.
- At least 3 years of relevant experience in computer architecture, CPU/GPU performance, kernel optimization, or high-performance computing.
- Experience working with CPU/GPU simulators.
- Exposure to performance profiling and debugging on system pipelines.
- Comfort with C++ and Python.
- Strong analytical and problem-solving skills.
Benefits
- Opportunity to build an AI platform beyond GPU constraints.
- Opportunity to publish and open-source cutting-edge AI research.
- Opportunity to work on a high-performance AI supercomputer.
- Job stability with startup vitality.
- Non-corporate work culture with continuous learning, growth, and support.
- Equal-opportunity workplace committed to diversity and inclusion.
Categories
About Cerebras Systems
Cerebras Systems designs and sells AI compute systems built around its wafer-scale WSE-3 processor, delivered as the CS-3 appliance and via the Cerebras Cloud. It targets enterprises, model labs, and government users needing fast training and inference, and offers on‑prem and cloud deployments. Privately held and headquartered in Sunnyvale, California, the company announced a multi-year partnership with OpenAI to deploy large-scale inference capacity.