Research Engineer, GPU Performance

Join Mistral, a leading AI solutions provider, as a Research Engineer specializing in GPU performance. In this role, you will develop low-level code to maximize GPU capacity, rethink model architectures for efficient inference, and integrate optimizations into a high-level MLOps framework. You will also profile and benchmark large-scale training and inference on thousands of GPUs. Ideal candidates will have expertise in writing custom low-level kernels, distributed computation infrastructure, and a strong understanding of generative AI.

Responsibilities

• Develop low-level code that takes full advantage of high-end GPUs and maximizes their capacity, including custom kernels and memory-layout optimizations.
• Rethink parts of the generative model architecture to make them more suitable for efficient inference, in collaboration with Research Scientists.
• Integrate low-level efficient code into a high-level MLOps framework, ensuring that optimizations benefit every training run and deployment.