Inference Platform Engineer — Scalable LLM Serving

Mistral; Greater London; Híbrido; GBP 120 000 - 160 000; Tempo integral

Resumo da oferta

Mistral is hiring for an Inference Foundation engineer to own the core of the inference stack, from the engine to orchestration, and to deliver production-grade serving for frontier models.

This hybrid role covers production LLM serving, engine and platform development, and capacity engineering. You will tackle scalable optimization, elastic capacity, and high-throughput training pipelines while collaborating across distributed teams.

Qualificações

• Experience building and running ML/LLM services at scale with latency targets.
• Hands-on experience with inference engines such as vLLM, sgLang, TensorRT-LLM.
• Familiarity with distributed serving architectures and GPU networking.

Responsabilidades

• Develop and fix the core of the inference stack — engine and orchestrator — including feature selection, configuration, and tuning for maximum performance at scale
• Own the release process for the serving stack: validated, regression-free releases through automated performance gates and progressive rollout
• Drive improvements and fixes upstream when the open-source engine is the right place for them
• Drive optimization of serving efficiency and topology at scale—reducing startup times, cache misses, and data movement

Conhecimentos

ML/LLM services at scale

Inference engines

Distributed serving architectures

CUDA/NCCL debugging

Python backend tooling

Kubernetes

GPU networking basics

Rust or C++ production

Ferramentas

PyTorch

CUDA

NCCL

Docker

Kubernetes

Nsight Systems/Compute

Profiling tools

Descrição da oferta de emprego

Mistral is hiring for an Inference Foundation engineer to own the core of the inference stack, from the engine to orchestration, and to deliver production-grade serving for frontier models.

This hybrid role covers production LLM serving, engine and platform development, and capacity engineering. You will tackle scalable optimization, elastic capacity, and high-throughput training pipelines while collaborating across distributed teams.