A collection of projects born from time pressure, wild constraints, and that specific kind of focus you only get when the clock is running. Some broke. Most shipped. All taught something.
Everything here was built under hackathon conditions — a hard deadline, a single idea, and whatever tools were at hand.
A prompt engineering assistant that rewrites vague inputs into structured, model-ready prompts. Built to solve the "I don't know how to ask" problem.
Dead-simple QR code creation that runs entirely in the browser — no uploads, no server, no fuss. Born from needing one in the middle of a talk.
Browser-native format conversion — JPEG, PNG, WebP, and more. No file uploads, no privacy trade-offs. All processing happens locally.
A beer colour calculator for homebrewers. A genuinely niche idea that turned into a surprisingly useful tool for a very specific audience.
A fast-paced multiplication quiz with a score streak mechanic. Proof that even basic maths practice can get competitive when the UX is right.
The workbench is never empty. Whatever's currently half-built will land here — follow along on YouTube to catch it as it happens.
Every build follows the same rough structure — a constraint, a clock, and a commitment to shipping something real.
One problem, one constraint, one hour to scope it down to something that could actually ship. No product roadmaps, no consensus meetings.
Core functionality only. Get the thing working before making it pretty. AI assists heavily here — pair programming on steroids.
Polish the edges that matter, cut anything that doesn't. This is where scope creep fights for its life and usually loses.
Done is better than perfect. The build goes live, gets documented, and earns its place in the tools drawer or the graveyard.
A 24-hour window stops you overthinking. You build the essential version, learn what the core actually is, and end up with something cleaner than a month of planning would produce.
Every "done" build is a reference point. You learn more from something that's live and getting used than from something polished in a dev environment for six months.
One person, one day, a useful tool. That wasn't realistic before. AI assistants compress the implementation gap — the idea-to-ship distance is shorter than it's ever been.