Blog

I got tired of finance apps that required accounts, sync, and cloud storage for something that should just live on my phone. So I built one — a complete tracker in a single HTML file. No npm. No server. Works offline. Forever.

Campus WiFi blocks most outbound ports. Cloud LLM APIs cost money per token. So I built a local RAG system that reads research papers and answers questions about them — running entirely on a mini PC under my desk, zero cloud.

A farmer with a smartphone shouldn't need an agronomist to know if their crop is sick. I trained MobileNetV2 on 87K leaf images to identify 38 plant diseases — 96.4% accuracy, under 2 seconds per prediction, deployed live.