{"repo":"kawsher-hridoy/DIU-AI-Mentor","free":true,"listed":false,"github":"https://github.com/kawsher-hridoy/DIU-AI-Mentor","clone":"git clone https://github.com/kawsher-hridoy/DIU-AI-Mentor.git","description":"AI-powered university mentoring: a deterministic 8-signal risk engine ranks every mentee, an LLM narrates — briefings, summaries, meeting prep, grounded chat. FastAPI + Next.js 14.","language":"Python","stars":11,"topics":["ai","education","explainable-ai","fastapi","mentoring","nextjs","student-success"],"license":null,"category":"auth-billing-email","readme_excerpt":"AI Mentor A risk-ranked command center for university student mentoring — deterministic scoring, AI narration, no black box. Built for the DIU AI Project Competition 2026 by team Cortex Crew . FastAPI · SQLAlchemy 2.0 · Next.js 14 · TypeScript · Tailwind · Azure OpenAI --- The problem A university mentor is typically responsible for 50+ students. The existing mentoring portal shows each student's results, attendance, payments, and meeting history — but only as flat tables, one student at a time, across 15 different screens . Nothing answers the question a mentor actually has: \"Which of my students needs me today , and why?\" So struggling students are noticed late, often after a failing semester — when a dropping attendance curve had been visible in the data for months. The idea AI Mentor sits on top of the existing portal (same API shapes, read-only) and turns its raw data into decisions, in two strictly separated steps: 1. A deterministic risk engine ranks — every student gets a 0–100 risk score computed from 8 weighted signals. Pure arithmetic: auditable, reproducible, unit-tested. The contributor breakdown is shown in the UI and always sums to exactly the score. 2. An LLM narrates — Azure OpenAI (or a built-in offline mock) turns the already-computed numbers into plain language: a daily briefing, a per-student summary, meeting prep, grounded Q&A chat. The model never invents, changes, or influences a score. That separation is the core design decision. The ranking can be de","default_branch":null,"files":null,"tree":[],"storefront":"/r/kawsher-hridoy","claimed":false,"request_supported":{"post":"https://gitbuyer.com/r/kawsher-hridoy/DIU-AI-Mentor/request-supported","requests":0},"note":"indexed from public GitHub; nothing is for sale on this page. Clone it from GitHub. Paid listings live at /search."}