Set up and harden one Ubuntu VPS, install two local GGUF models, written handoff

I need one Linux VPS set up, locked down, and loaded with two local AI model files. This is not a website. Not an app. Not a public chatbot. Do not install WordPress, cPanel, CyberPanel, Plesk, mail, Docker swarms, or anything I did not list. I will buy the VPS myself before work starts. Hostinger or DigitalOcean. Ubuntu twenty-four point zero four. About sixteen gigabytes of RAM if the plan allows, eight if that is all the budget holds. I send you login after we agree. Do this only: 1. Create a non-root sudo user. 2. SSH keys only. Disable password login. Disable root SSH. 3. Firewall. Open only SSH and any one port we agree on. 4. Tell me the exact DNS record for one subdomain I name. I enter it. You do not get my domain account. 5. Install Ollama, or llama.cpp if you have a clear reason. 6. Download only two GGUF files I name after hire. Confirm both load. 7. Bind the API to localhost, or put a password in front of it. No open public URL. 8. Write a one-page handoff: how I log in, what you installed, file paths, how I start and stop the models, what I should change after you leave. 9. Remove your keys and access when I confirm the handoff. Do not: • Transfer any domain • Ask for GoDaddy, Cloudflare, Apple, or email passwords • Pick different models • Add a pretty web UI • “Tune,” fine-tune, or customize the models • Keep access after handoff I will send the two exact filenames in private after hire. They are ordinary public GGUF files, about two and a half gigabytes and about four and a half gigabytes. NOTE: this is a one time job. It is unlikely there will be further work. (the iOS app has a glitch, and I cannot mark the appropriate box.)