Curriculum Builder Agent

Teachers spend 6+ hours every week planning curriculum — writing lesson plans, designing quizzes, sequencing difficulty, and adapting for students with different learning needs. Most of this time goes into applying pedagogical frameworks like Bloom's Taxonomy and Universal Design for Learning, expertise that not every teacher has time to develop. Curriculum Builder Agent solves this with a multi-step AI reasoning pipeline. A teacher enters a topic, grade level, difficulty, and optional accessibility needs (dyslexia, ADHD, ELL, visual/hearing impairment). In under a minute, the agent generates: 5 learning objectives mapped to Bloom's Taxonomy levels (Remember → Create) 5 detailed 50-minute lesson plans using the 5E instructional model 15 quiz questions (3 per day) with explanations, aligned to each day's cognitive level 5 hands-on activities with curated resources A weekly capstone project with a grading rubric A difficulty progression chart grounded in Cognitive Load Theory The pipeline runs Step 1 (objectives) sequentially, then Steps 2–5 (lesson plans, quizzes, activities, progression) in parallel using Promise.allSettled, cutting generation time roughly in half. Every step is grounded using Microsoft Foundry IQ, which retrieves pedagogical knowledge (Bloom's Taxonomy, Vygotsky's ZPD, Sweller's Cognitive Load Theory, CAST's UDL guidelines) and injects it into each prompt — reducing hallucination on educational content and satisfying the hackathon's Microsoft IQ integration requirement. The system is a full-stack MVC application: a React frontend communicating with a Node.js/Express REST API via Server-Sent Events for real-time generation progress. Teachers can export the curriculum as a PDF, view a print-ready A4 layout, take the quizzes interactively, and revisit previously generated curricula from auto-saved history. Built solo in four days for the Reasoning Agents track, targeting Best Reasoning Agent, Best Use of IQ Tools, and Accessibility awards.