Site Reliability Engineer, On-premise AI_FedGPT

FedGPT is Taiwan AILabs' in-house enterprise-grade on-premises LLM deployment platform, enabling government agencies, healthcare institutions, financial organizations, and academic bodies to adopt generative AI in fully air-gapped environments. Our customers include medical centers, research universities, and public sector organizations that demand the highest standards of system stability, reproducibility, and security.

【About the Role】
This role sits between our engineering team and the real-world environments where FedGPT runs. You'll be hands-on with deployment — from preparing GPU servers in-house, to installing them at customer sites, to troubleshooting issues when they come up. When you're not in the field, you'll work with our platform team on improving our internal tooling, CI/CD, and deployment automation.
It's a good fit if you have a solid Linux and Kubernetes foundation and want to broaden your experience across hardware, AI infrastructure, and customer-facing engineering — or if you've been doing field engineering and want to grow into platform / SRE work.

【Responsibilities】
1. Deploy FedGPT to customer sites:Prepare GPU servers in-house: install Ubuntu, NVIDIA drivers, CUDA, Kubernetes, and FedGPT. Run QA acceptance scripts. Deliver and commission systems at customer facilities, including networking and certificate setup.
2.Execute bi-annual upgrades in air-gapped environments:Travel to customer sites to run upgrade scripts, including DB migrations. Troubleshoot environment differences when they appear during the upgrade window.
3.Support customer issues as the first responder:Triage incoming issues, collect logs, and diagnose problems across hardware, OS, Kubernetes, and the application layer. Escalate to engineering oncall when needed.
4.Improve our deployment and platform tooling:Between field work, contribute to our platform team's roadmap — improving GitLab CI/CD, deployment scripts, and the dev/stg/prd environments. Help turn the manual parts of our work into automation.

【Essential Qualifications】
1.Hands-on Linux administration (Ubuntu)
2.Working knowledge of Kubernetes — comfortable debugging pods, services, and basic networking
3.Shell scripting (Bash) and at least one scripting language (Python preferred)
4.Familiar with Git and basic CI/CD concepts
5.3+ years of experience in infrastructure, DevOps, system administration, or related hands-on roles

【Preferred Qualifications】
1.Experience with NVIDIA GPU drivers, CUDA, or AI/ML infrastructure
2.Familiarity with Helm, ArgoCD, or GitLab CI/CD
3.Experience with database migrations (PostgreSQL )
4.Experience deploying or operating LLM serving stacks (vLLM, etc.)
5.Background working with enterprise customers or in regulated industries
6.Experience writing runbooks or training materials

【What We Offer】
1.Broad exposure. You'll touch hardware, AI infrastructure, Kubernetes, customer environments, and platform engineering — a combination that's hard to find in any single role.
2.Real systems, real users. Your deployments go to production environments that matter, in some of Taiwan's most demanding enterprise settings.
3.Growth path. This role is designed as a stepping stone toward Solution Architect, SRE, or Platform Engineering tracks within Taiwan AI Labs.