GadgetDoc AI: I Turned My Daily Repair Diagnosis Into an AI Engineer — Built Entirely From My Android Phone

#devchallenge #weekendchallenge

DEV Weekend Challenge: Passion Edition Submission

This is a submission for the DEV Weekend Challenge: Passion Edition.

What I Built

GadgetDoc AI — an AI diagnostic assistant for phones, laptops, and gadgets.

Describe a symptom — "phone won't charge," "laptop overheats after 10 minutes," "screen flickers when I open camera" — and it reasons through the problem like a senior repair engineer would: ranked likely causes with confidence scores, safe DIY steps you can actually try, honest red-flag warnings when a fault is dangerous enough to see a professional, and a realistic repair cost estimate in Naira.

🔗 *Live app:

*https://gadget-doc-ai-zpo6.vercel.app

Why This Is My Passion Project, Not Just a Hackathon Idea

I run Hilarious TechHub, reselling and repairing phones for students in Nigeria, and I'm studying Mechanical Engineering. Diagnosing what's actually wrong with a device — before a customer spends money on the wrong fix — is something I do in my head every single day. This weekend I decided to encode that instinct into software instead of just carrying it around as intuition.

The goal isn't a toy demo. It's something my own customers could genuinely use to walk into a repair shop already informed, instead of trusting whatever the technician tells them.

How I Built It

Design system — a deliberate "precision instrument" aesthetic instead of a chatbot look: warm white background, brass-gold accents, Fraunces for the headline, Space Grotesk for UI text, and IBM Plex Mono for every diagnostic number — confidence scores, cost ranges — so data reads like an instrument readout, not decoration.

Frontend — React + Tailwind CSS, a two-panel "workbench" layout (diagnostic form on the left, live console on the right) built on a CSS grid that stacks vertically on mobile.

AI — powered by Google Gemini (Google AI). The model receives the device type and symptom description and returns structured JSON: ranked causes with confidence percentages, DIY steps, red flags, and a Naira cost bracket — which the console renders directly into the report card.

The differentiator — every line of this, frontend, prompt design, and deployment, was built entirely on an Android phone (Redmi 14C) using Termux and hosted on vercel. No laptop, no traditional IDE. It's the same mobile-first workflow I use to run my whole tech business.