
PredictoR is a solo-built, AI-enhanced prediction market platform designed for the Qubic Track. It allows users to place bets using QUBIC tokens on real-world events, with outcomes auto-resolved by a Groq-hosted LLaMA 3 model, serving as a decentralized oracle and eliminating manual outcome resolution or central authority bias. The system comprises: Frontend: A user-friendly React interface to browse predictions, place bets, and view outcomes. Backend: An Express + TypeScript API integration with Groq AI, triggering event evaluations through secure prompts and parsing responses. Smart Contract: C++ logic compiled into Qubic EFI, handling user registration, betting, payout calculations, and event status, ready for deployment. While I built and tested all layers locally, including generating Qubic.efi with full contract logic, time constraints prevented a final QEMU runtime test. Regardless, PredictoR proves end-to-end feasibility—AI-led resolution with blockchain-native execution, all built alone from scratch. This project highlights combining AI with smart contracts, showing how trustless automation can elevate prediction markets. With more time, I’ll nail the final runtime deployment. Built for Raise Your Hack 2025.
8 Jul 2025