The opportunity
Cerebras Systems builds the world's largest AI chip, 56 times larger than GPUs. This architecture allows Cerebras to deliver industry-leading training and inference speeds; over 10 times faster than GPU-based hyperscale cloud inference services.
What you'll do
Post-training and reinforcement learning: Techniques used to improve model deployment quality through further training, tuning, RL, and focus on particular downstream tasks;
Dataset curation and optimization: Techniques to collect and select high-quality data, which can help models to train or tune more quickly or to higher quality;
LLM Pretraining: Techniques to ensure stability and compute-efficiency while pretraining high quality models. May include training dynamics, parameterizations, numerics, or others;
Sparsity : Techniques to sparsify models or data that improve training: time-to-quality, or optimize inference speed or throughput;
Domains : Coding agents, reasoning agents, generative language, image, video.
Build up a team capable of industry research and advanced development.
What they're looking for
- Organize various advanced development topics into cohesive agenda.
- Adapt novel algorithms and model architectures to run on the Cerebras platform.
- Systematically train, tune, and evaluate models to guide/advise production scenarios.
- Collaborate with other teams to co-design next-generation hardware and software architectures.
