Inference Engine Engineer - Scalable AI Serving

Mistral; Warszawa; Híbrido; PLN 180 000 - 300 000; Tempo integral

Resumo da oferta

Mistral is seeking an experienced engineer to join the Inference Foundation team in a hybrid role spanning production LLM serving, engine and platform development, and capacity engineering. You will tackle three challenges: scaling the inference stack, elastic capacity, and training the frontier models with high-performance serving.

You will own core stack development, optimize throughput and latency, and manage release processes through automated gates.

Qualificações

• Experience building and running ML/LLM services at scale with latency and availability targets.
• Hands-on experience with inference engines such as vLLM, SGLang, TensorRT-LLM.
• Familiarity with distributed serving architectures; debugging across the full stack (CUDA/NCCL, containers, networking, storage).

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.
• Optimize serving efficiency across the fleet — driving down pod startup time, tackling cold-cache regressions on scale-up, smarter caching and offloading.
• Build the serving infrastructure that powers RL and post-training for frontier models.

Conhecimentos

ML/LLM services at scale

Python

Kubernetes

CUDA/NCCL

Distributed serving

Ferramentas

vLLM

SGLang

TensorRT-LLM

CUDA

NCCL

Descrição da oferta de emprego

Mistral is seeking an experienced engineer to join the Inference Foundation team in a hybrid role spanning production LLM serving, engine and platform development, and capacity engineering. You will tackle three challenges: scaling the inference stack, elastic capacity, and training the frontier models with high-performance serving.

You will own core stack development, optimize throughput and latency, and manage release processes through automated gates.