Every platform in the hackathon ecosystem serves organizers. Devpost manages registrations. TAIKAI handles judging. HackerEarth runs pipelines. None of them are open when the clock starts and a team of four is trying to figure out which database to use.
Codefest.ai is the tool that was missing. The curated resource layer. The workspace that gets teams aligned before they write a line of code. The place where the work from last year’s hackathon still lives, forkable and learnable from.
It was built by Evren Arat — a hackathon participant, not a hackathon vendor — out of the specific frustration of wasting the first hour every time.
The point was never the prize. It was what happened to you inside 48 hours of real-stakes building — the discovery, the collaboration, the 2am moment when something finally works. The prize was just proof that it happened. We build everything around that experience, not around the leaderboard.
Nobody became a better engineer by independently discovering which npm package handles authentication. That friction was overhead. It was never teaching anything. The learning was always in the building, the scoping, the teamwork, the problem — and those things will always be entirely beyond our reach, by design.
Every component in our library is a link to a public GitHub repo. We don't redistribute code. We don't build anything for participants. We add the context that Google search can't: setup times vetted against real hackathon conditions, difficulty ratings that account for team size and time pressure, compatibility notes from projects that actually shipped.
AI is flattening the barrier to building. A team with no prior experience can ship something genuinely impressive in 24 hours now. That changes what hackathons are actually testing — and it means the differentiator is moving up the stack, toward problem selection, domain understanding, creative synthesis. That's where it should have always been. We're positioned at that inflection point.
Only about 7% of hackathon projects show any activity six months after the event. That's a structural failure, not a talent failure. The teams did the work. The ideas were real. The ecosystem just never built a way to preserve them, fork them, or learn from them. We're building that layer — so the work compounds instead of disappearing.
Faculty run hackathons because they believe something transformative happens in that room. We believe that too. Our job is to protect it — by removing the friction that was never part of the design, not by redefining the experience. If we ever become something that makes organizers uncomfortable, we want to hear why. We'd probably agree.
These aren’t disclaimers. They’re decisions. The things we won’t build are as important as the things we will.
Esports created speedrunning infrastructure — leaderboards, replay systems, community-vetted routes, shared techniques — that made competition something you could study, improve at, and build a community around. Hackathons have nothing equivalent. Every team starts from scratch. Every winning stack disappears after demo day. Every cohort reinvents the same solutions.
That’s the gap Codefest is closing. Not just the first hour — but the entire infrastructure layer that was always missing: the preserved knowledge, the curated patterns, the institutional memory that makes each hackathon build on the last one rather than starting over.
AI is accelerating this. When the execution barrier flattens, the differentiator becomes the quality of the question you’re asking. The problem statement. The domain understanding. The creative synthesis that no tool can generate for you.
Codefest is positioned exactly at that inflection point: handling the infrastructure so participants can focus on the part that matters — and building the preservation layer so the work that gets done in those 48 hours doesn’t vanish on Monday morning.
One day truly innovate. Build something that matters. Come out different on the other side. That’s what this is for.