Independent DevOps & Infrastructure Engineer GCloud-Lab / AgentForge
I build and operate a multi-application cloud-native AI platform on GKE for multi-tenant AI agent hosting, real-time trading microservices, and an autonomous news-quant pipeline. The platform is fully Git-driven and has more than 1,000 commits.
I implement zero-touch GitOps with Terraform and Flux CD so infrastructure and application changes are audited and synchronized from Git. I operate vLLM serving for Qwen models up to 27B parameters on tiered NVIDIA GPU capacity, including L4, RTX 6000, and A100 nodes.
I reduced GPU costs by approximately 70% through KEDA-driven scale-to-zero workloads on spot GPU nodes. I also implemented tenant-isolated CloudNative-PG databases, Cilium NetworkPolicies, SOPS and Age secret encryption, Tailscale-only internal access, CVE scanning, and automated remediation pipelines.