GPU Performance Research Engineer - Scale LLMs

Mistral; Paris; On-site; EUR 120,000 - 180,000; Full time

Job summary

Mistral AI is seeking a Research Engineer focused on GPU performance. You will work with Research Scientists to push frontier models training and inference efficiency, writing low-level code and rethinking architectures to scale.

Join a distributed team across regions, contributing to high-end hardware utilization and end-to-end model deployment with a strong emphasis on collaboration and practical impact.

Qualifications

• Experience writing custom low-level kernels for GPUs and optimizing throughput.
• Familiarity with large-scale distributed GPU clusters and interconnects.
• Strong understanding of generative AI, LLMs, and model fine-tuning for deployment.
• Solid software design skills, testing, code review, and CI/CD practices.
• Self-starter with a collaborative, low-ego working style.

Responsibilities

• Develop low-level code that maximizes GPU capacity—from kernels to memory layouts and communication optimizations.
• Rethink model architectures for efficient inference in collaboration with researchers.
• Integrate low-level optimisations into a high-level MLOps framework to benefit multiple runs.
• Profile, benchmark, and squeeze performance on large-scale training and inference across GPUs.
• Model LLM execution on various GPUs to assess hardware usage and build serving estimates.

Skills

Low-level kernels

GPU clusters

Generative AI

CI/CD

Collaborative

Job description

Mistral AI is seeking a Research Engineer focused on GPU performance. You will work with Research Scientists to push frontier models training and inference efficiency, writing low-level code and rethinking architectures to scale.

Join a distributed team across regions, contributing to high-end hardware utilization and end-to-end model deployment with a strong emphasis on collaboration and practical impact.