The projects below show how Munish approaches infrastructure work: define the operational problem, automate the repeatable parts, measure the result, and document the trade-offs.
PortfolioMe
The site you are reading is built with Hugo and deployed through Cloudflare Pages. Docker, Nginx, Kubernetes, Helm, Kustomize, and Gateway API packaging remain in the repository as local and proof-of-work deployment layers.
- PortfolioMe case study
- Source repository
- Docker image optimization article
- Kubernetes cost optimization article
Self-hosted AI inference platform
A k3s and NVIDIA GPU platform running Ollama and Open WebUI, with GPU runtime configuration, MetalLB, persistent storage, and GitOps-style deployment practices.
Production troubleshooting and automation
The technical blog documents Kubernetes health checks, OOMKilled diagnosis, Azure workload authentication, Docker dependency failures, and resource-cost analysis from hands-on infrastructure work.