KidExplain
Local-first web app using an LLM to translate concepts into age-appropriate language for children.
A personal project: a local-first web app that uses an LLM to translate any concept into age-appropriate language for children — type a question like “Why is the sky blue?” and get back an explanation tuned to the child’s age group. Runs entirely on-device via Ollama, so no data ever leaves the machine.
Features
- Age-aware explanations across three bands: baby talk (0–2), simple words + analogies (3–5), short explanations (5–8)
- Multilingual: English, Chinese (Simplified), and Farsi
- Local & private by default (Ollama), with an OpenAI-compatible API mode for swapping in a cloud model
- Responsive, mobile-friendly layout with local-network serving for use on phones/tablets
Architecture
A FastAPI backend exposes a single POST /api/explain endpoint that builds an age- and language-aware prompt and routes it to either a local Ollama model (gemma4:e2b) or an external OpenAI-compatible API, returning the explanation as JSON. The React/Vite/TypeScript frontend handles age-group and language selection and renders the response.
Tech stack: React, Vite, TypeScript, Tailwind CSS, FastAPI, Ollama
GitHub: [GITHUB LINK TBD]