A gully cricket umpire you can shout at

Developer Abhijit has created 'Gully Umpire,' an AI-powered tool designed to automate scoring for street cricket matches in India. Built as a Next.js application, the tool uses a pipeline of open-weight models to handle match logistics. Whisper large v3 turbo transcribes vocal calls, while GPT-OSS and Qwen models interpret complex, messy cricket terminology—including Hinglish—to update the scorecard. The system is designed for reliability, using a deterministic code-based engine to track runs and wickets, ensuring that the AI only handles natural language processing rather than core logic. The project, which emphasizes open innovation, utilizes Groq for free model hosting and Render for deployment. By allowing users to define custom lane rules and providing an undo feature for noisy environments, the app aims to solve the common issue of scorekeepers being distracted from the game, effectively bringing AI utility to casual outdoor sports.