NeuroVerse: An AI-Driven Peer-to-Peer Learning and Mentorship Platform

Most online learning systems deliver course material well, but they do little to connect a learner with a suitable mentor, check what the learner already knows, or give people a reason to return after the first few sessions. This paper presents NeuroVerse, a web platform for peer-to-peer learning in which every user can both teach and learn. Learners take a skill assessment, are matched to mentors by a transparent weighted score, and book paid sessions using in-platform credits, with an AI study assistant available between sessions. The platform uses Next.js, React and Tailwind CSS on the front end, Next.js API routes running on Node.js, and a PostgreSQL database hosted on NeonDB. Accounts are protected with JSON Web Tokens stored in HTTP-only cookies and bcryptjs password hashing. Assessment questions are drawn at random from a local bank and graded by exact match for multiple-choice items and by a keyword heuristic for code, and a threshold on the percentage score assigns a beginner, intermediate or advanced level. Mentors are ranked on a 100-point score built from skill, level, language, rating and availability. A Gemini 1.5 Flash model powers the study assistant, Jitsi Meet links host the sessions, and daily riddles with credit rewards encourage repeat visits. The prototype was checked manually, and no automated tests or user study have been carried out. Credit purchases are simulated, submitted code is not executed, and several production features, such as real payments, mentor availability and access control, remain to be built.