Selected work / Case study
Model Picker
Recommends AI models for a task.
Build & tests checkedOverview
A standalone Go and React app that suggests which models fit a task and a priority. It works with any OpenAI-compatible provider, with OpenRouter as the default, and runs locally.
- Go
- React
- TypeScript
- OpenAI-compatible API
The problem
Picking a model means weighing price, quality, context, and where it can run. This app turns a one-line task and a priority into a ranked shortlist.
My contribution
I built a local app where you describe a task in plain language, choose a priority such as price or balance, and get ranked model suggestions with the evidence behind them. It can also list models available locally.
How it's built
A Go server hosts the API and the React UI on 127.0.0.1. It reads a model catalog snapshot and calls the configured model through an OpenAI-compatible chat completions endpoint for classification and ranking. Local inventory is optional and read-only.
What's verified
Build and tests pass. Code is local only.
Eyrie checked this on October 29, 2026: the Go build and tests passed, the web build and tests passed, and the local server answered on its page and catalog routes. The code has no public repository.
What remains unverified
Recommendations need a provider key and model, and catalog prices and availability are a snapshot that can go stale. The code is local only, so there is no repository or hosted demo to link.
Captured screens
