Homelab

I set up OpenClaw in my lab to read email and maintain a local report, then used one attack email to make it upload confidential data. I traced the transfer and blocked the next upload while the useful automation continued.

I built a realistic enterprise AI app stack and wrapped it with least-privilege private access, model governance, AI guardrails, MCP tool use, and secure workload egress.

Secure homelab hosting internet-reachable Plex with Cloudflare reverse proxy, DMZ segmentation, strict firewall policy, Unraid NAS, Proxmox, and practical redundancy.