AI Project
WorldEd
A full-stack educational platform that transforms STEM learning through intelligent course delivery, real-time progress tracking, and an AI learning assistant.
Next.js 15 · React 18 · PostgreSQL · Gemini 1.5 Flash · Drizzle ORM
GitHub01 / Overview
WorldEd is the platform I founded and run: a full-stack STEM learning system with LaTeX-rendered content, embedded video lectures, interactive quizzes, and a gamified streak system — built on Next.js 15, TypeScript, PostgreSQL, and Drizzle ORM.
The headline engineering wins are measurable: React Server Components cut client-side JavaScript by 40%, optimised composite indexes cut query latency by 40%, and database-level aggregation reduced API traffic by roughly 70%.
At its core is Eddi, a Socratic AI tutor powered by a RAG pipeline over pgvector with Google Gemini function calling — students ask in natural language and Eddi guides them through the material instead of handing over answers.
02 / Demo
03 / Key Features
Eddi — Socratic AI Tutor
RAG pipeline over pgvector with Gemini function calling. Students query in natural language and Eddi navigates them through course material in a closed-loop agent workflow.
Dynamic Progress & Streaks
Real-time progress tracking calculated server-side, with a gamified, timezone-aware daily streak system.
Rich Academic Content
Full LaTeX rendering, syntax highlighting, and a structured hierarchy (Course → Module → Quiz) with JSONB schema-free questions.
Enterprise-Grade Security
Clerk RBAC auth, rate limiting, CORS whitelisting, and rotating JWTs.