Work

AI-driven, multi-cluster Kubernetes management and monitoring platform inspired by Rancher, with a LiteLLM + RAG diagnostic engine that reads logs and suggests automated fixes.

• Kubernetes
• FastAPI
• LiteLLM
• RAG
• Helm

Agentic AI grocery planning engine deployed on Azure. Provisioned Container Apps infrastructure with Terraform, secured production with Private Links and Managed Identities, and traced flows with Application Insights.

• Azure
• Terraform
• Gemini API
• LangGraph

Self-hosted LLM inference on a Kubernetes cluster bootstrapped with kubeadm across a master and worker nodes. The KubeRay operator watches a RayService CRD and reconciles a RayCluster of head and worker pods, with Ray Serve deployments hosting the models and scaling replicas across the workers.

• kubeadm
• KubeRay
• Ray Serve
• Kubernetes
• LLM Inference