itxjoshy/smart-home-energy-ai

AI-powered smart home energy management app for West African households # smart-home-energy-ai Energy is expensive in Nigeria. Like, genuinely expensive — and most people have no idea where it's all going. You get your NEPA bill at the end of the month and it's just a number. No breakdown. No context. No way to know if the AC is killing you or if it's the freezer you leave running all night. I built this to fix that. --- ## What it does Connects to your smart plugs, watches your power usage over time, and tells you what's actually eating your electricity — and what you can do about it. The suggestions come out in plain English, calculated in Naira, and it works fine on a bad connection. It also tracks your NEPA supply separately from your generator. Because in Nigeria those are two very different things and pretending otherwise doesn't help anyone. --- ## Why I built it I'm from Abuja. I've watched my family pay high electricity bills for years without understanding why. Every month the bill arrives and every month it's a surprise. I wanted to build something that gives ordinary households actual visibility — not charts for the sake of charts, but information that helps you make a decision. Most energy apps are built for the UK or US. This one isn't. --- ## Stack - **Frontend** — React + JavaScript (Vite) - **Styling** — CSS - **Backend** — Firebase Cloud Functions - **Database** — Firebase Firestore (real-time) - **Smart plug data** — Mock data generator (simulates Nigerian household usage patterns) - **ML** — TensorFlow.js for usage prediction - **Recommendations** — LLM API for natural language suggestions - **Deployment** — Vercel --- ## How it works ``` mock data generator → data pipeline → firestore → tfjs model → llm → dashboard ``` The mock data generator simulates realistic Nigerian household power consumption — NEPA load shedding schedules, generator switching, and appliance patterns included. The TensorFlow.js model learns usage patterns over time and predicts what consumption will look like over the next day …