DEV Community: Ayush Kasaudhan

GhostMemory AI — Memories That Refuse to Disappear

Ayush Kasaudhan — Thu, 29 Oct 2026 16:18:18 +0000

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

What I Built

I built GhostMemory AI, a browser-based AI storytelling application that gives forgotten memories a second life.

Users can write a personal memory, story, person, place, or moment and choose how they want the AI to transform it:

👻 Ghost Story
🕯️ Gentle Remembrance
✨ Poetic Memory
💌 Letter From The Past

I interpreted the "Ghost in the Machine" theme as the idea of memories becoming echoes inside a machine.

The "ghost" doesn't have to be supernatural. It can be a person, a place, a story, or a moment from the past that we don't want to disappear.

GhostMemory AI takes those memories and turns them into something that can be read and remembered again.

Demo

🌐 Live Website:

https://ayush-kasaudhan-01.github.io/-GhostMemory-AI/

💻 GitHub Repository:

https://github.com/ayush-kasaudhan-01/-GhostMemory-AI

Note: The file:///C:/... URL only works on my own computer, so I used GitHub Pages for the public demo.

Code

The complete source code is available on GitHub:

https://github.com/ayush-kasaudhan-01/-GhostMemory-AI

The project is built using:

HTML
CSS
JavaScript
Transformers.js
SmolLM2-135M-ONNX
Browser LocalStorage

How I Built It

The core of GhostMemory AI is an open-weight language model running in the browser.

I used SmolLM2-135M-ONNX with Transformers.js for text generation.

The basic flow is:

User Memory
      ↓
GhostMemory AI Interface
      ↓
Prompt Builder
      ↓
Transformers.js
      ↓
SmolLM2 Open-Weight Model
      ↓
Generated Memory
      ↓
Local Browser Storage

There is no commercial AI API involved.

The application sends the user's memory to the locally running model in the browser, which generates the response.

Generated memories are stored using browser localStorage, so the project doesn't require a database or backend server.

The interface was built from scratch using HTML, CSS, and JavaScript.

Why Does Open Innovation Matter?

For a project dealing with personal memories, privacy matters.

A closed AI API would mean sending someone's memories to an external service.

With an open-weight model running locally in the browser, the project can keep the AI interaction much closer to the user's own device.

Open innovation also gives the project more freedom.

I can:

🔒 Keep personal memories away from a commercial AI API
💰 Avoid per-request AI API costs
🔄 Replace the model with another compatible open model
🧠 Change the prompts and AI behavior
🛠️ Inspect and modify the implementation
🌐 Build without depending on a proprietary AI provider

This was one of the biggest reasons I wanted the AI itself to be an open component rather than simply connecting a website to a closed API.

My Agent Session

I did not use an external agent session for this project.

The project was built around a local/browser-based open-weight AI workflow instead.

The Spirit of It

The idea behind GhostMemory AI is simple:

What if a machine could help us keep the things we are afraid to forget?

A small memory can disappear from a conversation, a notebook can be lost, and a story can stop being told.

But when that memory is transformed into a story, it becomes another way of preserving it.

That is where the "ghost" in GhostMemory AI comes from.

The machine isn't bringing back a supernatural ghost.

It is bringing back an echo of something that already happened.

Prize Categories

I'm entering this project for the categories relevant to:

🤖 Open-source / open-weight AI
👻 Ghost in the Machine
🌐 Browser-based AI
🔒 Privacy-focused AI

Final Thoughts

GhostMemory AI started as a simple idea: combine the atmosphere of Halloween with the human need to remember.

Building it with open AI made the idea even more interesting because the "machine" doesn't have to be a distant server.

It can be right there with you.

Some memories disappear.

Some become ghosts.

GhostMemory AI gives them a place to stay. 👻

🚀 SellerSpark AI: An Open-Source AI Listing Generator for Online Sellers

Ayush Kasaudhan — Thu, 15 Oct 2026 09:25:01 +0000

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle

What I Built

I built SellerSpark AI, a privacy-first AI product listing generator for small online sellers.

Writing product listings can be time-consuming, especially for small businesses and individual sellers. SellerSpark AI allows users to enter simple product details and generates a complete, marketplace-ready listing.

It generates:

🏷️ Product titles
📝 Product descriptions
✨ Selling points
🔍 Search keywords
#️⃣ Hashtags

The target users are small online sellers, marketplace sellers, handmade-product businesses, and social-media sellers who want to create professional product listings faster.

A potential customer would pay for this because it can save time and reduce the effort required to create optimized product content for multiple products.

Demo

🌐 Live Demo: https://ayush-kasaudhan-01.github.io/SellerSpark-AI/

The basic workflow is:

Enter Product Details → Generate Listing → Copy & Use

The AI runs directly in the browser, so users don't need to set up a backend or provide an API key.

Code

💻 GitHub Repository: https://github.com/ayush-kasaudhan-01/SellerSpark-AI.git

The complete source code is available in the repository.

How I Built It

SellerSpark AI is built around an open-weight AI model running directly in the browser.

AI Stack

SmolLM2 — open-weight language model
Transformers.js — browser-based AI inference
ONNX — model execution
JavaScript — application logic
HTML & CSS — frontend and user interface

The application takes the seller's product information and creates a structured prompt for the open-weight model.

The model then generates the product title, description, selling points, keywords, and hashtags.

One of the main goals was to avoid depending on a traditional closed AI API. The project works as a browser-based AI application without requiring a backend server or API key.

Why Does Open Innovation Matter?

Open innovation made it possible for me to build an AI-powered side-hustle product without relying completely on a paid closed AI API.

Using an open-weight model provides:

🔒 Better privacy
💰 Lower operating costs
🔄 Freedom to experiment with different models
🧩 More control over the AI workflow
🚫 Less vendor lock-in
🌐 The possibility of running inference directly in the browser

For small businesses and independent developers, these advantages can make AI tools more accessible and affordable.

SellerSpark AI is an example of how open AI technology can be turned into a practical product that solves a real business problem.

Prize Categories

Side Hustle
Open-Source AI

TouchGrass AI — An Open-Weight AI That Gets You Off the Screen and Into the Real World

Ayush Kasaudhan — Fri, 09 Oct 2026 11:11:45 +0000

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass

What I Built

I built TouchGrass AI, an AI-powered outdoor adventure companion designed to turn screen time into real-world activity.

The idea is simple: instead of keeping users inside an AI chat, the AI gives them a reason to put the phone down and go outside.

Users choose how much time they have, their environment, their mood, and what kind of experience they want. TouchGrass AI then generates a personalized outdoor mission, such as exploring nature, taking a short walk, observing plants or birds, or completing a small outdoor challenge.

The website also includes:

🤖 AI-generated outdoor missions
🌿 Different environments and adventure styles
⏱️ Active mission mode with a timer
📸 AI-powered nature/photo discovery
⭐ XP and progress tracking
📖 Outdoor adventure journal
🌙 Dark mode
📱 Responsive design for mobile and desktop

The target audience is anyone who spends too much time on screens and needs a simple push to get outside.

The core philosophy is:

The best AI interaction is the one that gets you to stop using the AI.

Demo

Live Website:
https://ayushkasaudhan957-eng.github.io/TouchGrass-AI/

The website is deployed and can be used directly in the browser.

Code

GitHub Repository:
https://github.com/ayushkasaudhan957-eng/TouchGrass-AI

The complete source code is open and available in the repository.

How I Built It

TouchGrass AI is built around open-weight AI models accessed through Hugging Face Inference Providers.

For text generation, I used:

OpenAI GPT-OSS-120B (openai/gpt-oss-120b)
Qwen2.5-VL-3B-Instruct (Qwen/Qwen2.5-VL-3B-Instruct) for vision-based image analysis

Hugging Face Inference Providers provides access to open models through an OpenAI-compatible chat-completion endpoint, allowing the application to use hosted inference without requiring users to download a local model.

Architecture

User
  ↓
TouchGrass AI Website
  ↓
HTML + CSS + JavaScript
  ↓
Node.js / Express Backend
  ↓
Hugging Face Inference Providers
  ↓
Open-Weight AI Models
  ↓
Personalized Outdoor Mission / Nature Analysis

The frontend handles the user experience, mission builder, timer, XP, journal and image upload.

The backend acts as a secure AI proxy so that the Hugging Face API token is not exposed in the browser.

For mission generation, the user's preferences are converted into a structured prompt and sent to the open-weight language model.

For the nature discovery feature, an uploaded image is sent to a vision-language model which analyzes what is visible and provides an AI-generated observation.

I intentionally chose hosted inference instead of requiring users to install Ollama, LM Studio, or download large models. This makes the project much easier to try while still keeping open-weight models at the core of the AI functionality.

Why Does Open Innovation Matter?

Open innovation makes this project possible because the AI layer is not loc