Madarbulb: a private remembrance builder with Gemma 4 running in your browser

Md Shahbaz Hashmi Ansari — Sat, 31 Oct 2026 18:03:54 +0000

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 4: Ghost in the Machine

What I Built

Madarbulb is a private Día de los Muertos remembrance builder. You enter a loved one’s name and a memory, with optional details like their favorite food, song, and photo. Gemma 4 writes a short remembrance and a gentle calaverita, then the app presents them on a candlelit virtual ofrenda decorated with marigolds, papel picado, candles, and offerings.

I interpreted “Ghost in the Machine” as a way to remember people with care while keeping their stories and photos close. The model helps shape a tribute, but it writes about the person in the third person and only uses details the user shares.

Demo

Demo: https://ofrenda-rho.vercel.app/

Code

Repository: https://github.com/ShahbazCoder1/-OFRENDA

How I Built It

Madarbulb is a Next.js App Router app built with TypeScript and Tailwind CSS. It runs Gemma 4 E4B through LiteRT-LM’s browser API using WebGPU. Instead of downloading the model from the app, the user selects the .litertlm file already on their device. The browser passes that file directly to LiteRT-LM; it isn’t uploaded or copied into app storage.

Each generation uses a fresh conversation and a separate prompt. The first streams a 90–140-word remembrance; the second streams a 6–8-line affectionate calaverita. Both prompts tell the model not to impersonate the deceased or invent details. The app also supports English and Spanish, resizes an optional portrait in the browser, saves ofrendas in localStorage, and exports the finished altar as text or PNG.

Why Does Open Innovation Matter?

These stories and photographs are personal. Local inference means they don’t need to be sent to a hosted AI service. There’s no account, API key, or per-generation inference charge, and the project can run without internet access after the app and model file are available on the device.

Using an open model also gives the project room to change. The model can be swapped for another compatible option, such as a smaller Gemma variant, based on the device. A closed hosted API could provide text generation, but would require trusting another service with private memories and photos.

The Spirit of It

The altar makes the generated words feel connected to the details around them: a portrait, a favorite song, or a dish someone remembers. The model doesn’t speak as the person who died. It helps the living make a small space for remembering them.

Thik Achhe? Building an offline, multilingual scam checker for my friend's grandfather

Md Shahbaz Hashmi Ansari — Sun, 04 Oct 2026 18:15:08 +0000

This is a submission for the Hacktoberfest Weekend Challenge: Build for a Friend

What I Built

My friend's grandfather retired a few years ago and now spends a lot of time on his phone and the family desktop. Recently, he received a terrifying SMS claiming his bank account would be blocked by evening unless he clicked a link to complete his KYC. He almost called the number. His family keeps telling him not to click suspicious links, but he is genuinely scared of ignoring a real bank message. He also rightly refuses to paste his messages into random internet chatbots because they contain real bank account numbers and OTPs.

To solve this, I built "Thik Achhe?" (which translates to "Is this okay?"). It is a simple web page that can be installed on his phone or desktop. He just pastes the message, taps one massive button, and gets a clear Red, Yellow, or Green answer in his native language (it supports English, Hindi, Bengali, and Nepali). It gives him one simple line on why it was flagged, and one line on what he should do next.

Demo

You can try it live here: https://thik-achhe.devloper.xyz/

Watch a quick demo of how it works: https://youtube.com/shorts/0lnH-yV-_rI

Code

You can find the full source code in my GitHub repository here: https://github.com/ShahbazCoder1/Thik-Achhe

How I Built It

I built the application using Next.js (App Router), Tailwind CSS, and TypeScript.

The core of the project relies on the open-weight gemma-4-E2B-it-web.litertlm model. Instead of hosting the model on a server, I integrated Google's @mediapipe/tasks-genai library to run the LLM inference entirely inside the client's browser.

The application uses a two-step verification process:

A hardcoded TypeScript rules engine instantly checks the text for known scam patterns like URL shorteners, panic keywords ("blocked within 24 hours"), or explicit asks for an OTP.
The message is then passed to the local Gemma model with a strict prompt demanding a structured JSON response in the user's selected language.

If the rules engine finds a hard red flag, it permanently locks the verdict to "Red" to ensure the AI can never accidentally hallucinate and mark a dangerous message as safe.

Why Does Open Innovation Matter?

Open innovation was absolutely critical for this project. If I had to rely on a closed API, I would be taking a retired senior citizen's most sensitive text messages (containing real one time passwords and bank digits) and beaming them to a third-party server.

By utilizing open-weight models like Gemma and client-side inference tools like MediaPipe, his messages never leave his physical device. It is incredibly fast, perfectly secure, and costs absolutely nothing to host or maintain. A closed system would have made this project either a massive privacy violation or financially unsustainable.

Prize Categories

Best Use of Gemma

DEV Community: Md Shahbaz Hashmi Ansari

Madarbulb: a private remembrance builder with Gemma 4 running in your browser

What I Built

Demo

Code

How I Built It

Why Does Open Innovation Matter?

The Spirit of It

Thik Achhe? Building an offline, multilingual scam checker for my friend's grandfather

What I Built

Demo

Code

How I Built It

Why Does Open Innovation Matter?

Prize Categories