Cloud Orchestration Engineer for Scalable ML Inference

Inferact; United States; Remote; USD 140,000 - 190,000; Full time

Benefits offered by this job

Visa sponsorship

Fully remote

Job summary

Inferact is seeking a cloud orchestration engineer to build the operational backbone that keeps vLLM running reliably at scale. You’ll design the systems for cluster management, deployment automation, and production monitoring that enable teams worldwide to serve AI models without friction.

You’ll ensure that vLLM deployments are observable, debuggable, and recoverable, turning operational complexity into infrastructure that just works.

Qualifications

• Bachelor’s degree or equivalent experience in computer science, engineering, or similar.
• Strong experience with Kubernetes and container orchestration at scale.
• Experience designing and implementing custom Kubernetes operators.
• Proficiency in Python/Rust/Go and infrastructure-as-code tools (Terraform, Helm, etc).
• Experience managing GPU clusters and debugging hardware issues.
• Ability to work across cloud platforms (AWS, GCP, Azure) and on-premise infrastructure.

Responsibilities

• Design cluster management, deployment automation, and production monitoring for large-scale environments.
• Ensure deployments are observable, debuggable, and recoverable.
• Collaborate with teams across the globe to keep vLLM running reliably at scale.
• Turn operational complexity into reliable infrastructure and streamlined workflows.

Skills

Kubernetes

Container orchestration at scale

Custom Kubernetes operators

Python/Rust/Go

GPU cluster management

Multi-cloud & on-prem infrastructure

Education

Bachelor’s degree or equivalent

Tools

Terraform

Helm

Job description

Inferact is seeking a cloud orchestration engineer to build the operational backbone that keeps vLLM running reliably at scale. You’ll design the systems for cluster management, deployment automation, and production monitoring that enable teams worldwide to serve AI models without friction.

You’ll ensure that vLLM deployments are observable, debuggable, and recoverable, turning operational complexity into infrastructure that just works.