AI Inference Platform Engineer: GPU Ops & Reliability

KRAFTON; United States; On-site; USD 150,000 - 210,000; Full time

Job summary

KRAFTON is hiring a senior ML infrastructure engineer to lead the operation and evolution of large-scale inference services at our US-based office. You’ll own the serving platform, GPUs, and the control plane, collaborating with AI research and product teams to deliver reliable, scalable AI capabilities.

You will design deployment automation, observability, and cost-aware GPU utilization, while advancing SLOs and incident response culture. Remote options may apply depending on project needs.

Qualifications

• Engineering experience: 5+ years with 3+ years in infra/SRE or ML systems.
• Experience deploying on Kubernetes with load balancing, scheduling, and auto-scaling.
• Build observability via metrics/logs/traces and handle on-call and postmortems.
• IoC/GitOps/CI/CD to reproduce infra and deployments.
• Ability to document decisions and collaborate across teams.

Responsibilities

• Operate large-scale inference services and improve the serving platform.
• Build and run serving platform and control plane on Kubernetes GPU clusters.
• Automate onboarding and deployment with canary/rollback; manage API gateway auth, quotas, and rate limits.
• Observe latency, throughput, availability, and error rates; analyze bottlenecks and failures.
• Manage GPU resources and unit cost measurement for workloads.
• Improve routing, scheduling, caching, and model integration in serving logic.
• Validate optimization ideas (quantization, speculative decoding) in production.

Skills

ML systems

Kubernetes

SRE

IaC/GitOps

Performance tuning

Postmortems

Tools

vLLM

SGLang

TensorRT-LLM

Triton Inference Server

Job description

KRAFTON is hiring a senior ML infrastructure engineer to lead the operation and evolution of large-scale inference services at our US-based office. You’ll own the serving platform, GPUs, and the control plane, collaborating with AI research and product teams to deliver reliable, scalable AI capabilities.

You will design deployment automation, observability, and cost-aware GPU utilization, while advancing SLOs and incident response culture. Remote options may apply depending on project needs.