Site reliability engineer in inference platform

Nebius; United States; Remote; USD 120,000 - 180,000; Full time

Benefits offered by this job

Competitive compensation

Career growth

Flexibility and ownership

Collaborative culture

Impactful AI projects

International environment

Job summary

Nebius Group ищет инженера инфраструктуры для обеспечения надёжности и производительности нашего инференс-стека в условиях больших нагрузок. Ваша роль — проектировать и поддерживать системы мониторинга, делать так, чтобы сервисы восстанавливливались автоматически и масштабировались под GPU-требования.

Требуется экспертиза в Kubernetes, Prometheus, Grafana и Terraform, опыт с IaC, а также умение писать на Python или Bash и работать над SLA задачи.

Qualifications

• Опыт работы с продакшн-системами на распределённых бэкэндах и управление ими.
• Навыки проектирования систем мониторинга, трассировки и алертов.
• Опыт работы с IaC и инфраструктурой как код.

Responsibilities

• Обеспечение надёжности, производительности и наблюдаемости всей инферент-стека.
• Проектирование и refinement телеметрических пайплайнов — метрики, логи, трассировки.
• Настройка авто масштабирования Kubernetes для эффективного использования GPU.
• Написание Terraform-модулей для устойчивости кластера.
• Укрепление маршрутизации запросов и обработки сбоев, чтобы пользователи не замечали проблем.
• Использование автоматизации и runbooks для быстрого обнаружения и устранения инцидентов.
• Содействие культуре после инцидентов для предотвращения повторений.
• Обеспечение масштабирования платформы при достижении целей по надёжности и затратам.

Skills

Kubernetes

Prometheus

Grafana

Terraform

Infrastructure as Code

Python / Bash

SLOs

Distributed systems

Debugging performance

Collaboration

Tools

Kubernetes

Prometheus

Grafana

Terraform

Job description

Описание

Token Factory is a part of Nebius Cloud, one of the world’s largest GPU clouds, running tens of thousands of GPUs. We are building an inference platform that makes every kind of foundation model — text, vision, audio, and emerging multimodal architectures — fast, reliable, and effortless to deploy at massive scale. To deliver on that promise, we need an engineer who can make the platform behave flawlessly under extreme load and recover gracefully when the unexpected happens.

Задачи

• own the reliability, performance, and observability of the entire inference stack.
• designing and refining telemetry pipelines — metrics, logs, and traces that turn hundreds of terabytes of signal into clear, actionable insight.
• tune Kubernetes autoscalers to squeeze more efficiency out of GPUs
• craft Terraform modules that bake resilience into every new cluster
• harden our request-routing and retry logic so even transient failures go unnoticed by users.
• rely on the automation and runbooks you helped create to detect, isolate, and remediate problems in minutes
• drive the post-mortem culture that prevents recurrence.
• scaling the platform smoothly while hitting aggressive cost and reliability targets.

Требования

• deep fluency with Kubernetes, Prometheus, Grafana, Terraform, and the craft of infrastructure-as-code.
• You script comfortably in Python or Bash
• understand the nuances of alert design and SLOs for high-throughput APIs
• have spent enough time in production to know how distributed back-ends fail in the real world.
• you care about building self-healing systems
• thrive on debugging performance from kernel to application layer
• enjoy collaborating with software engineers to turn reliability into a feature users never have to think about.
• Будет плюсом: Experience shepherding GPU-heavy workloads — whether with vLLM, Triton, Ray, or another accelerator stack — will serve you well; a background in MLOps or model-hosting platforms

Условия

• Competitive compensation
• Career growth and learning opportunities
• Flexibility and ownership
• Collaborative and innovative culture
• Opportunity to work on impactful AI projects
• International environment and talented teams
• Fast moving
• Bold thinking
• Constant growth
• Meaningful impact
• Trust and real ownership
• Opportunity to shape the future of AI