EMAIL:mensah-gbekor@hotmail.com PASSWORD: 0020101370
AI Kid Tutor – Project Story
What Inspired Us
The inspiration for AI Kid Tutor came during the shift to remote learning, when we noticed how many children—especially those aged 5 to 12—struggled with getting personalized support and attention. Most digital learning tools were either too generic or not age-appropriate, and few incorporated true interactivity or emotional intelligence. We envisioned a solution that could act like a caring, always-available tutor: responsive, safe, engaging, and tailored to each child’s learning journey.
How We Built It
We designed and built AI Kid Tutor as a full-stack, production-grade platform over several months, focusing on real-time performance, safety, and scalability. •Frontend: React 18 + TypeScript, Tailwind CSS, Vite for fast dev •Backend: Supabase (PostgreSQL) with Row Level Security and real-time subscriptions •AI System: Dual-provider system — OpenAI o1-mini (primary), Gemini 2.0 Flash (fallback), plus offline content •Features: Child-friendly UI, secure auth, gamified lessons, real-time analytics, parental controls
We structured the database with 15 interconnected tables to manage users, quizzes, games, learning sessions, recommendations, and more. Every interaction—be it a quiz result or game achievement—is tracked and analyzed to inform AI-generated feedback and personalized learning paths.
What We Learned •Designing for kids requires more than simplicity—it demands emotional safety, feedback, and fun. •Parents want transparency and control; our parent dashboard became essential. •We learned how to balance flexibility and control in AI outputs to prevent inappropriate or confusing responses. •Building for real-time, mobile-first performance while keeping costs low meant heavy optimization and smart caching.
Challenges We Faced 1. Voice recognition for kids: Whisper ASR needed tuning to understand young, high-pitched voices more accurately. 2. Data safety and privacy: Ensuring COPPA-friendly data isolation via Row Level Security added backend complexity. 3. Maintaining uptime: We built a three-layer AI fallback system to ensure reliability even if OpenAI or Gemini services went down. 4. Engagement retention: Kids lose focus fast—getting the right mix of game mechanics and learning took many iterations.
Log in or sign up for Devpost to join the conversation.