V.Vardhan / Pro

PERSONAL ENGINEERING LAB

Systems I build.
Problems I debug.
Lessons I document.

A running record of personal infrastructure work: self-hosted Kubernetes, Linux operations, networking, observability and experiments toward AI platform engineering.

I use a personal lab to explore what happens beyond the happy path: nodes change, DNS gets overwritten, containers end up attached to the wrong daemon, and applications need ingress, metrics and consistent operational workflows.

These are personal lab projects and learning notes. Architecture diagrams are simplified for publication; internal addresses and sensitive configuration are omitted. Professional work is documented separately. These pages describe documented activities, not a live service-status inventory.

6 stories · 4 learning / planned tracks

Documented personal work

Implemented

Kubernetes / Platform Engineering / Virtualization

Building a Self-Hosted Kubernetes Platform with Proxmox and K3s

A small multi-node personal K3s environment, with explicit ingress and service-IP decisions and a separate monitoring layer.

Proxmox VEK3sHelmNGINX IngressMetalLBLinuxTailscale

2 min read · Personal lab

Implemented

Observability / Kubernetes

Making a Homelab Kubernetes Cluster Observable

Deploying kube-prometheus-stack and accessing Grafana through NGINX Ingress to inspect a personal K3s cluster.

PrometheusGrafanakube-prometheus-stackHelmK3sNGINX Ingress

2 min read · Personal lab

FUTURE WORK · SEPARATE FROM IMPLEMENTED PROJECTS

Learning tracks and planned experiments

Milestones are proposals, not completed deployments or measured results. There are no case-study or repository links for unbuilt work.

Learning / research

AI platform and local coding agents

Exploring open-weight models, hardware sizing, quantization, serving concepts and coding-agent workflows. No working inference endpoint or GPU deployment is established.

  • Planned: choose and deploy one isolated model endpoint; record real hardware, runtime and setup.
  • Planned: publish a benchmark method before measuring latency, time to first token and throughput.
  • Planned: add authentication, logging, metrics and resource limits; compare one configuration change.

Planned

GitOps-controlled K3s deployment

A proposed demo application managed by Argo CD or Flux. Professional Argo CD familiarity does not establish a completed personal GitOps deployment.

  • Planned: implement and verify reconciliation, rollback and bootstrap before publishing a case study.

Planned

Home Assistant presence automation

A possible ESP32-C3 and LD2410-series presence prototype with ESPHome/Home Assistant. A finished physical prototype is not established.

  • Planned: define a low-voltage prototype and measure actual reliability. Any mains work requires appropriately certified equipment and professional installation.

Planned

Private homelab service inventory

Explore a secure way to document service ownership, backups and access. No running inventory application is claimed.

  • Planned: design a private inventory without publishing addresses, passwords or secret references.

What counts as a project here?

A deployed service, an operational problem with a documented resolution, or an explicitly labeled experiment. Each page states what was actually done and what remains to be built.

Read professional Selected Work separately →