AI Engineer, Model Quality and Performance

About the role

Cerebras Systems builds the world's largest AI chip, delivering training and inference speeds over 10 times faster than GPU-based cloud services. As an AI Engineer, Model Quality and Performance, you will own model quality and performance for Cerebras' inference offerings: define what "good" looks like across the models served, build AI-driven systems to measure it at scale, and turn those signals into artifacts used by customers and the product team. You will sit between engineering, product and customer-facing teams.

Key responsibilities

• Design eval suites with AI agents in the loop, curating a mix of advanced, basic, long-context and customer-use-case-specific tests for each model release, using Claude to generate, validate and prune test cases.
• Build custom evals for target customers by having AI agents mine trajectories from their workloads and synthesize representative eval sets.
• Automate eval execution end-to-end with AI-driven pipelines on top of standard tooling (Docker, Git, CI), so the system runs itself between releases.
• Build automations that forecast and benchmark model performance on Cerebras for top customers, including estimating how fast customer workloads will run in production.
• Build product-quality tooling that combines quality and performance data into a single, easy-to-use view.

Required profile

• Experience building AI agents, shipping real systems with Claude (or equivalent) as a force multiplier.
• Strong math and statistics background.
• Taste for tooling design: something a non-engineer can use without complaining; AI helping you ship it is a bonus.

Required skills

• Comfort with Docker, Git and the standard automation stack.
• Bonus: performance-tuning experience on custom silicon, GPUs or FPGAs.
• Bonus: experience designing evals for agentic, coding, long-context or multimodal use cases.
• Bonus: familiarity with open-source eval frameworks such as EvalScope and lm-eval-harness.