Same thinking, applied to my own problems.
Side projects are product decisions, not coding exercises. Same thinking as the day job, different constraints. Every project here started from a real need: understand the problem, build the minimum to validate, decide what not to build. The result is on Telegram, not in a portfolio.
CasaHunter
ActiveProcesses 1,000+ apartment listings daily across 8 Italian platforms. Scores quality with Claude Sonnet. Telegram bot + dashboard. Zero competitors do this.
Python 3.11SQLiteClaude SonnetReact 18TailwindLeafletRechartsTelegram APIGitHub Actions
What it taught me
View project →Three-pass scoring: eliminate 50% deterministically (free), spend AI budget on the 50% that matters (€0.25/user/month). Same logic applicable to any product at scale.
If you're curious about how I think when I build, the professional case studies tell the same story at a different scale.