Popcorn Time: never burn popcorn again

Everyone has burned microwave popcorn. The instructions on the bag are a lie, the 'popcorn button' on the microwave is a bigger lie, and the rule of thumb — 'stop when pops are 2-3 seconds apart' — requires you to stand in the kitchen listening like a weirdo.

So in 2019 I made my first app out of it. You put your phone next to the microwave, start the app, and it listens to the popping. A machine learning model figures out the moment the popcorn is actually done — not the bag's guess, not the microwave's guess, the real moment — and alerts you. Perfect popcorn, no smell of regret in the kitchen for three days.

The honest origin story

I built it to learn iOS development and on-device ML with a problem small enough to finish. The pitch wrote itself: let thousands of scientists and millions of dollars of machine learning research do the legwork for your snack. It's a silly idea executed seriously, which is still my favorite genre of software.

The model listens for the pattern of pops, not just the gaps. Kernels popping is a surprisingly rich signal — the rate ramps up, peaks, and decays, and the decay tells you more than any timer.

What it taught me

Everything I've shipped since has the same shape: a tiny, specific annoyance, an opinionated solution, no feature creep. Popcorn Time has 65 ratings and a 4.3 average, which for a joke-adjacent app from 2019 I'll take.

My mom thinks it was a good idea to make this app. It's on the App Store — don't let her be wrong.