Writing
Everything I publish, newest first. Use the filters to narrow by article type or topic.
13 articles
How I Benchmark a Local LLM
The protocol, the commands, and the first QuietWatts row: Qwen3.8 27B on a DGX Spark, measured by a harness rather than a stopwatch on a chat window.
Can You Run AI on a NAS?
Usually yes, and it will be slower than you want. But the electricity is already paid for, which changes the argument more than the speed does.
Which Local Model Should You Run?
Your hardware sets a ceiling and the arithmetic can tell you exactly where it is. Which model to pick under that ceiling is a judgment nobody can make from a table, including me.
Run Your First Local LLM with Ollama, Step by Step
Install Ollama, pull a model that fits your hardware, and chat with a local LLM in about fifteen minutes, then the handful of commands that make it usable afterwards.
What a 32k Context Window Costs You in Memory
Context is not free and it is not fixed. The cache grows with every token you exchange, which is why a model that loaded fine runs out of memory an hour into a conversation.
How Loud Is Two Machines in One Room?
Adding a second machine adds three decibels, not double the noise, and that one fact changes where the money should go when a lab gets too loud.
NAS RAID Levels Explained for Newcomers
What each RAID level costs you in capacity, what it actually protects against, and why the rebuild after a failure is the part worth planning for.
Proxmox vs. TrueNAS vs. Unraid
Three systems built around three different assumptions about what a home server is for. Picking well means knowing which assumption matches yours, not which is best.
Quantization Explained (GGUF, AWQ, and Friends)
What quantization does to a model, why a four-bit file is never four bits, and the thing it buys you that almost nobody mentions.
What Size UPS Does a Home Lab Need?
Size on volt-amps, not on minutes, because the runtime you need arrives free and the rating on the box is not watts.
Cloud GPU vs. Owning Hardware: The Real Cost Math
What owning the hardware actually costs against paying per token, worked out at three usage levels, including the cases where buying never pays back.
What a Homelab Really Costs in Electricity
The arithmetic behind a homelab power bill, worked out from one number you can hold in your head, plus where the watts actually hide and which levers move the total.
How Much VRAM You Need for Every Model Size
The memory arithmetic for running a local model, worked out for every common size and quantisation level, including the term most guides leave out.