MLOps Engineer Hybrid

Публикатор: Наталья Обсуждение: @devops_jobs #vacancy #вакансия #sre #DevOps #MLOps Позиция: Mlops Формат: гибрид(релок,Франция или Узбекистан) Вилка: от 3000$ - 10000$на руки, трудоустройство по ТК страны Контакт для связи: @tokhastik_hr Задачи: - Own inference infrastructure end-to-end: optimize latency, throughput, and cost across our model fleet. - Build and scale model serving with TensorZero, vLLM/SGlang/TRT, and Kubernetes. - Design and maintain vector search pipelines with Vector storages. - Familiarity with support metrics (SLAs, FCR, deflection) and ability to define service health KPIs. - Turn research into product: grab experimental models from the research team, figure out what's production-ready, and ship it - formatting, sampling parameters, deployment, the whole thing Требования (обязательные): - 3+ years shipping high performance ML systems in production, not just training notebooks - Deep hands-on experience with inference optimization - you've debugged latency spikes and know the difference between theoretical and real-world throughput - Comfortable across the stack: from CUDA kernels to Kubernetes manifests to Grafana dashboards A big plus: experience with Rust, custom Triton kernels, benchmarks - Английский язык от В2 будет плюсом (но команда русскоговорящая) По всем вопросам @tokhastik_hr