Articles on open-source tools, infra, and dev practices.
Pick the model before you write the prompt. Lesson 1 of a four-part series on working with AI — model tiers, context windows, tokens, and a decision checklist you can keep.
Same model, very different results — because of the prompt, not the model. Lesson 2 of a four-part series on working with AI: prompt anatomy, acceptance criteria, and standing project context.
A model that re-learns your project every session will quietly break its own conventions. Lesson 3 of a four-part series on working with AI: documentation, rules, and skills — and which layer each piece of guidance belongs in.
Less for the same result, measured on screen — not a percentage quoted from somewhere else. The last lesson in a four-part series on working with AI: baselines, levers, and guardrails.
No jargon, no assumed tech knowledge — what an AI model is, what an API key is for, and why someone would use two different AI tools together on purpose.
No jargon, no assumed tech knowledge — what 'scraping' job boards means, what 'auto-apply' actually does, and why a human still has to click submit.
No jargon, no assumed tech knowledge — what it means to build and 'deploy' an app, and what all those developer tool names (GitHub, Vercel, Supabase) actually do.
Scrape 15+ job boards, match by keyword and salary, and auto-apply with browser automation — no AI, no cloud services, no API keys. MIT-licensed and multi-profile from a single install.
Route bulk AI work to a self-hosted Hugging Face agent and reserve Claude for precision tasks — full quality, a fraction of the cost.
A practical guide to setting up Terminal, VS Code, GitHub, Claude Code, Vercel, and Supabase — from a simple blog with CMS to a full table management app. This guide walks you through each tool with real AI prompts you can drop in immediately.
