Deployed Live

Deployed Live - www.vibeforces.tech The thing I kept thinking about, walking into this hackathon, was how unrecognizable our hiring pipelines have become from the way we actually build software now. I write code with Claude every day. So does almost every engineer I know. Yet the interviews we still give look like 2018 — reverse a linked list, balance a binary tree, prove you can manually do the thing the model now does in two seconds. There are mass vibecoders now. There is no standardized way to train them. No way to test them. No way to rank them. Engineers who ship with Claude are an entirely different population from engineers who ship without — and the industry has no instrument for telling them apart. I built VibeForces to be that instrument. A competitive assessment platform with Codeforces-style ratings, public leaderboards, live contests, and a recruiter test builder that lets companies grade candidates on how well they drive a model toward working software, not on how many algorithms they memorized. There are categories for spec-to-prompt translation, token-efficient golfing, bug-fix prompting, architecture ranking, and UI reproduction. Then there's the part of the platform I'm proudest of: a senior track where the candidate doesn't write a prompt — they drive an agent. Real distributed bugs reproduced and patched inside sandboxed sessions. Services built end-to-end. Agents designed, instantiated, and graded by their trace. That whole track lives inside Claude Managed Agents, because it couldn't have lived anywhere else. Three roles, full auth, deployed live at vibeforces.tech. You're not hiring someone who memorized merge sort anymore. You're hiring someone who can ship with Claude. VibeForces is where that question gets answered.