Code-Mentor

AI-Powered Code Review & Learning Platform

Graduation project — Faculty of Computers and Artificial Intelligence, Benha University · Class of 2026.

Code Mentor is an intelligent learning platform that simulates the experience of working under a senior developer. It combines adaptive skill assessment (IRT 2PL), AI-personalized learning paths (hybrid embedding recall + LLM rerank), and multi-agent code review (three parallel specialist agents) to bridge the gap between academic learning and real-world job requirements — delivering senior-level code feedback in under 5 minutes.

The platform ingests a learner's code submission (GitHub URL or ZIP), runs six static analyzers in parallel, then routes the code to an LLM review pipeline across six quality axes (correctness, readability, security, performance, design, task-fit). The output: a detailed report with line-by-line annotations, a RAG-powered Mentor Chat grounded in the learner's actual code, and a project-level audit feature for senior reviews of entire repositories.