mozaicdb/naviiq

AI career guidance agent for African students built with LangGraph and Qwen Cloud # Naviiq Naviiq is an AI-powered career guidance agent built for African students. It has a natural conversation with a student, learns about their background, interests, strengths, and goals, then generates a personalized career roadmap they can save, download, and share. This project was built for the **Global AI Hackathon with Qwen Cloud**, under **Track 4: Autopilot Agent**. ## Why Naviiq Many African students, especially in underserved communities, do not have access to proper career counseling. Naviiq fills that gap with an AI agent that asks the right questions, adapts to the student's age and context, and produces a roadmap that is specific to them rather than generic advice. The agent supports three conversation modes depending on the student's age: - **Explorer** (under 13): simple, encouraging, exploratory questions - **Discovery** (13 to 17): guidance focused on subjects, interests, and early direction - **Career** (18+): deeper conversation focused on career paths, skills, and concrete next steps ## How It Works Naviiq is powered by a 6-node LangGraph agent. Each node has a single responsibility, gathers information through conversation, and hands off to the next node until enough context is built to generate a roadmap. ``` collect_identity to collect_background to analyze_strengths to define_goals to decision_engine to roadmap_generator ``` - **collect_identity**: gathers name, age, and current school/work status - **collect_background**: explores academic background and areas of interest - **analyze_strengths**: identifies the student's natural strengths and working style - **define_goals**: understands what the student wants to achieve and their constraints - **decision_engine**: matches the student's profile against career paths and computes a confidence score - **roadmap_generator**: generates the final personalized roadmap once confidence is high enough (70% or above) Each node makes a single call to **Qwen3.7-Plus** via Qwen Cloud, usi …