/portfolio/portable-ollama

Portable Ollama

A zip-based local LLM setup that can run from a folder or USB drive without asking the host machine for a whole installation ritual.

projectPortable Ollama
statusportable build
surfacelocal AI packaging
index03 / 06
stack tagsOllama, LLMs, environment patching
OllamaLLMsenvironment patchingWindows-friendly ziplauncher scripts

What it was actually solving.

Portable Ollama was an attempt to make local LLMs less fussy to share: unzip the folder, run the launcher, and keep the host system as clean as possible.

Challenge

Ollama was not originally shaped around the exact portable workflow Calvin wanted. Turning somebody else’s software into a self-contained package meant learning which paths and environment assumptions mattered.

Solution

Patch the runtime environment around the folder the server was launched from, so model storage and supporting paths resolve locally instead of depending on a conventional machine-wide setup.

What stuck afterward

A good black-box modification is half patience and half observation. This project taught how to change the behavior of software without fully owning the codebase: trace assumptions, control the environment, and package the result so another person can use it.

Proof it actually moved.

A quick demo clip from the original project page, kept inside the same hard-edged terminal frame instead of burying it below the writeup.

demo source/static/portfolio/demo-videos/single-file-ollama-demo.mp4

More things from the same lab bench.

Same visual system as the main ToadTech revamp: hard-edged, terminal-ish, and allergic to generic portfolio gloss.