VPS, Linux, and networking tutorials

Short procedures, tested on our own machines in Montreal. No filler theory: the commands, the traps, and what to check before going to production.

10 articles RSS feed →

Server loading a language model and emitting tokens

Self-hosting a Hermes model on a VPS with Ollama

A quantized 8B model fits in 6 GB of RAM and answers at a few tokens per second with no GPU. Here is the honest build, numbers before commands.

Read the article → 11 min read

Agent loop connecting the model, its tools, and its observations

Running an AI agent 24/7 on a VPS

An agent looping on your laptop is not a service. Here is what to add so it survives a reboot, an API outage, and your invoice.

Read the article → 10 min read

Central MCP server exposing tools to a client

Exposing an MCP server on a VPS

MCP is the standard socket between an agent and your tools. A VPS is enough to expose one — as long as you don't leave the door open.

Read the article → 9 min read

Encrypted tunnel linking a client machine to a server

Setting up a WireGuard VPN on a VPS in 20 minutes

WireGuard fits in about a hundred lines of config and saturates a gigabit link on an 8.50 CAD VPS. Here is the full build, server and clients.

Read the article → 10 min read