DEV Community: S Ehtishamul Haq

Ghostlight 👻: Every Machine Has a Ghost Story | Open-Source AI Meets Halloween 🎃

S Ehtishamul Haq — Thu, 29 Oct 2026 04:37:19 +0000

Ghostlight: Every Machine Has a Ghost Story 👻

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

What I Built

I built Ghostlight, a spooky AI-powered story generator inspired by the theme Ghost in the Machine. The idea is simple: what if every machine had a story hidden inside it, waiting to be discovered?

Ghostlight allows users to enter a story idea, choose an atmosphere, and generate an eerie tale. Whether it's a forgotten arcade machine, a mysterious signal at midnight, or a room filled with old memories, Ghostlight transforms these prompts into atmospheric storytelling experiences.

The project offers two modes:

Local AI Mode: Uses Ollama to connect to an open-weight language model running locally.
Offline Mode: Uses a built-in JavaScript story generator, allowing users to create stories without installing an AI model.

I wanted to combine Halloween-inspired creativity with the principles of open-source AI, privacy, and accessibility.

Demo

👻 Live Website: https://ehtishamhaque1212-arch.github.io/Ghostlight/

Explore the interface, choose a story prompt, select your preferred atmosphere, and summon a story from the darkness.

Code

💻 GitHub Repository: https://github.com/ehtishamhaque1212-arch/Ghostlight.git

The project is built with HTML, CSS, and vanilla JavaScript, with optional integration with Ollama for local AI inference.

How I Built It

I built Ghostlight using:

HTML to structure the application and its interactive controls.
CSS to create a dark, atmospheric interface with purple accents and a responsive layout.
JavaScript to manage user inputs, generate offline stories, and communicate with the local AI service.
Ollama to run an open-weight language model locally and generate original stories without relying on a hosted AI API.

For local AI generation, Ghostlight sends a prompt to Ollama's local API, which returns a generated story. Users can customise the prompt, atmosphere, and story length.

I also implemented an offline fallback so the application remains usable without an AI model. This fallback is template-based rather than an LLM, but it makes the experience accessible even when local AI is unavailable.

Why Does Open Innovation Matter?

Open innovation matters because AI should be something developers can understand, experiment with, and adapt—not just consume through closed APIs.

With Ghostlight, using an open-weight model through Ollama gives users more control over model selection and allows story generation to happen locally. Prompts can remain on the user's computer instead of being sent to a remote AI provider through a hosted API.

This approach also provides flexibility. Developers can experiment with different compatible models, modify prompts, adjust generation settings, and improve the application without depending on a single proprietary AI service.

The offline mode adds another layer of accessibility by allowing users to explore the project without needing a model or a continuous internet connection.

For me, open-source AI makes experimentation more accessible and gives developers the freedom to build tools that reflect their own ideas.

My Agent Session

I used AI-assisted development to help plan and build the interface, implement the story-generation logic, and document the project.

The project code and implementation are available in the GitHub repository above.

Note: I haven't included a DevRelay session link because I don't have a session URL to share.

The Spirit of It

Ghostlight began with a simple thought: what if the machines we use every day could tell stories about the things we've forgotten?

A flickering screen, an abandoned arcade machine, or an old message appearing at midnight can all become the beginning of a mystery.

But not every ghost story needs to be frightening. Some stories are about memories, people, and moments we want to preserve. That is why I wanted Ghostlight to explore both eerie fiction and the emotional side of remembering.

This project is my small experiment in bringing open-source AI and creative storytelling together—one mysterious signal at a time.

Prize Categories

My project is intended to align with the open-weight AI / local inference theme through its optional Ollama integration.

I'll confirm the official partner prize categories against the challenge rules before selecting any specific partner category.

Built with curiosity, code, and a little bit of darkness. 👻

Project: Ghostlight

Live Demo: https://ehtishamhaque1212-arch.github.io/Ghostlight/

Source Code: https://github.com/ehtishamhaque1212-arch/Ghostlight.git

LazyFlow AI 🚀: Work Smarter, Not Harder with Open-Source AI

S Ehtishamul Haq — Sun, 25 Oct 2026 15:06:47 +0000

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 3: Slow and Steady.

🚀 LazyFlow AI: Work Smarter, Not Harder

What I Built

LazyFlow AI is an AI-powered productivity workspace designed to simplify everyday tasks and help people work smarter, not harder.

The idea is simple: spend less time organizing tasks and dealing with repetitive work, and more time focusing on what actually matters.

Built with HTML, CSS, and JavaScript, LazyFlow AI combines an interactive web interface with AI-assisted productivity features. It is designed for students, developers, and anyone looking for a simpler way to organize their workflow.

The project explores how open-source AI can make productivity tools more accessible without relying entirely on expensive, hosted AI APIs.

The mission: Less busywork. More meaningful work.

Demo 🌐

Try LazyFlow AI here:

👉 https://ehtishamhaque1212-arch.github.io/LazyFlow-AI/

The website is hosted on GitHub Pages, making it accessible directly from a browser.

Code 💻

GitHub Repository:

https://github.com/ehtishamhaque1212-arch/LazyFlow-AI

The project is organized into just four files:

index.html — Website structure and interface.
style.css — Styling, layout, and visual design.
script.js — Application logic, interactive features, and AI integration.
README.md — Project documentation and setup instructions.

Keeping the project in four files makes it easier for beginners and contributors to understand the code and get started.

How I Built It 🛠️

I built LazyFlow AI using three core web technologies:

HTML to structure the website.
CSS to create the interface and visual experience.
JavaScript to power interactions, application logic, and AI integration.

For the local AI setup, I chose Ollama with the Gemma 3 1B model as the intended inference environment.

This approach explores running an open-weight AI model locally instead of depending entirely on a paid, cloud-based AI API.

The project brings together a lightweight web interface, productivity functionality, and an approach to AI that developers can explore and customize.

My main priorities were simplicity, accessibility, and making the codebase easy to understand and extend.

Why Does Open Innovation Matter? 🌍

Open innovation makes AI experimentation more accessible.

Developers should be able to experiment with AI, learn how models work, and build useful applications without necessarily paying for hosted APIs or creating complex infrastructure from scratch.

Open-weight models and tools such as Ollama make this possible.

For LazyFlow AI, open innovation offers several important advantages:

Accessibility: Developers can experiment with supported models without requiring a paid hosted-model API.
Local control: Users can run compatible models on their own machines.
Privacy potential: When configured for local inference, prompts can stay on the user's machine rather than being sent to a third-party AI service.
Customization: Developers can experiment with models and adapt the application to their needs.
Community collaboration: Other developers can inspect the code, suggest improvements, and contribute new features.

Open innovation helps transform individual experiments into projects that others can learn from, improve, and build upon.

That is what makes open-source AI exciting: it gives more people the opportunity to become creators, not just consumers, of AI-powered technology.

My Agent Session 🤖

I don't have a public agent-session recording to share yet.

The Lazy Payoff ⏳

The goal of LazyFlow AI is to reduce the friction involved in managing everyday work.

By bringing productivity features and AI assistance into one workspace, the project aims to help users spend less time on repetitive tasks and more time on meaningful work.

I haven't conducted formal user testing to measure time savings yet, so I won't claim a specific number. The focus is on making everyday workflows simpler and exploring how local AI can support that goal.

Being lazy isn't about doing nothing. It's about finding smarter ways to get things done.

Prize Categories 🏆

Primary focus: Open-Source AI

LazyFlow AI explores the use of open-weight AI models and local inference in a practical web application.

Final Thoughts

LazyFlow AI is my contribution to exploring how open-source AI can make everyday productivity more accessible.

This project is a step toward building tools that are simple to use, easy to understand, and open to improvement.

I'm excited to keep learning, building, and improving through the open-source community.

Built with HTML, CSS, JavaScript, and a little less unnecessary effort. 🌱

Hacktoberfest #OpenSource #AI #JavaScript #WebDevelopment #Ollama #Gemma

🚀 HustleCraft: Your Open-Source AI Sidekick for Smarter Product Listings

S Ehtishamul Haq — Thu, 15 Oct 2026 09:53:22 +0000

HustleCraft - AI Product Listing Studio

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

What I Built

🚀 HustleCraft — AI Product Listing Studio

I built HustleCraft, a web application designed to help small business owners, online sellers, and independent creators turn basic product details into professional e-commerce listings.

Selling online is not just about having a great product. Sellers also need compelling titles, engaging descriptions, and relevant search keywords. Writing all of these manually takes time, especially for people running a side hustle alongside their studies, jobs, or other responsibilities.

HustleCraft aims to make this process easier.

✨ What it can do:

Generate product titles that are ready to refine for an online store.
Create engaging product descriptions.
Suggest relevant search tags and keywords.
Support different product details, audiences, and writing styles.
Provide copy-ready content that sellers can adapt for their listings.
Offer a demo mode and an optional integration with locally running AI through Ollama.

The goal is simple: help small sellers spend less time writing listings and more time growing their businesses.

Potential users include Etsy sellers, Instagram businesses, handmade-product creators, and small retailers who want an affordable way to improve their product listings.

Demo

🌐 Try HustleCraft live:

https://ehtishamhaque1212-arch.github.io/HustleCraft/

Explore the interface, enter product details, and try generating a listing.

Code

💻 GitHub Repository:

https://github.com/ehtishamhaque1212-arch/HustleCraft

The project is built with:

HTML — page structure and content.
CSS — responsive layout and visual design.
JavaScript — user interactions, listing generation, and application logic.
Ollama integration — designed to support locally running open-weight AI models.

I wanted to keep the project lightweight and accessible while making it possible to extend the AI capabilities.

How I Built It

I built HustleCraft using HTML, CSS, and JavaScript, with a focus on keeping the application simple to run and easy to understand.

The project includes a demo experience for testing the product-listing workflow and an optional local AI integration using Ollama with an open-weight model such as Llama 3.2.

The intended local AI workflow is straightforward:

The seller enters the product name, description, target audience, and preferred writing style.
The application prepares a prompt with the supplied product information.
A locally running Ollama model generates a title, description, and search tags.
The seller can review and copy the generated content.

The demo mode provides a way to explore the interface without downloading a model. It uses templates rather than generative AI, so the local model is what enables the actual AI-powered generation.

One important distinction: the deployed GitHub Pages website is a static frontend. Local AI generation requires Ollama running on the user's own computer and a compatible browser connection; it does not automatically run on GitHub Pages.

Why Does Open Innovation Matter?

For a side hustle, the cost of every tool matters.

A seller who needs to generate product listings regularly may not want another recurring subscription or a bill that grows with every API request. Open-weight models offer an alternative: run inference locally, experiment with different models, and choose how the system behaves.

Open innovation matters to HustleCraft in three main ways:

🔒 Privacy and control

With a properly configured local setup, product information and prompts can stay on the user's computer instead of being sent to a third-party AI provider. This can be particularly useful for sellers working with unpublished products or internal business information.

💸 Lower recurring AI costs

Local inference does not require a per-request fee from a hosted model API. There are still costs associated with hardware, electricity, and setup, but users can avoid usage-based API charges.

🧩 Freedom to experiment

Open-weight models make it possible to swap models, experiment with prompts, and adapt the generation process for different types of sellers without being locked into one provider's pricing or API.

For this project, open AI is not just an extra feature. It is the foundation of the local-generation approach.

My goal is to explore how open-source tools can make practical AI assistance more accessible to people building businesses on a small budget.

Prize Categories

HustleCraft is built around open-weight AI and local inference. I will select the applicable partner prize categories listed in the official challenge guidelines.

Final Thoughts

Building HustleCraft has been an opportunity to explore how open-source AI can solve a small but real problem for people running side hustles.

I want to keep improving it based on actual seller feedback, especially around listing quality, useful keywords, and how easily someone can turn a product idea into a listing they are happy to publish.

Better Listings. Bigger Opportunities. ✨

If you run a small business or sell products online, I'd love to hear what you would want an AI listing assistant to do next!

TerraAI: Build with Open-Source AI, Touch Grass, and Reconnect with Nature

S Ehtishamul Haq — Fri, 09 Oct 2026 11:14:01 +0000

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

🌿 What I Built

I built TerraAI — Your AI Outdoor Companion, a web application designed to encourage people to spend less time staring at screens and more time experiencing the world around them.

In a world where we spend hours on our devices, TerraAI takes a different approach: instead of keeping users online, it uses technology to inspire them to step outside.

🌱 What TerraAI offers:

AI-powered outdoor missions: Generate ideas for simple outdoor activities using a locally running AI model.
Mission library: Explore activities such as nature walks, mindful breaks, and neighborhood exploration.
Activity timer: Track time dedicated to outdoor activities.
Progress tracking: Earn points, record completed missions, and track outdoor minutes.
Persistent progress: Save statistics and checklist selections in the browser.
AI fallback: Continue exploring built-in missions when the AI service is unavailable.

TerraAI is built for students, people who spend a lot of time on their computers, and anyone looking for a simple way to take meaningful breaks.

The idea behind the project is simple:

Technology should help us experience more of the real world, not just spend more time looking at a screen.

🚀 Demo

🌐 Live Website: https://ehtishamhaque1212-arch.github.io/TerraAI/

Visit TerraAI, explore the available missions, start an activity timer, and track your progress.

Note: AI-generated missions require the local Ollama service to be configured and running. The built-in mission experience is available without it.

💻 Code

🔗 GitHub Repository: https://github.com/ehtishamhaque1212-arch/TerraAI

The project is built with HTML, CSS, and JavaScript, with optional local AI inference through Ollama.

The repository includes the website structure, responsive styling, application logic, mission generation workflow, timer, and browser-based progress storage.

I wanted to build something lightweight, understandable, and easy to extend without requiring a large frontend framework.

🛠️ How I Built It

I built TerraAI using:

HTML5 for the website structure.
CSS3 for the responsive layout and visual design.
JavaScript for mission selection, timers, progress tracking, and application interactions.
Ollama as the local AI runtime.
Llama 3.2 as the example open-weight model for generating outdoor activity suggestions.
Browser localStorage for saving user progress without requiring a cloud database.

The AI workflow is designed to run locally. When a user requests new missions, the application sends a prompt to the Ollama API, asks the model to return structured mission suggestions, and displays the results in the interface.

If the AI request fails, TerraAI falls back to predefined missions so the core experience remains usable.

This project helped me explore how open-weight models and local inference can be incorporated into a practical web application.

🔓 Why Does Open Innovation Matter?

For TerraAI, open innovation matters because the purpose of the project is to make AI useful without making a remote AI service a requirement for the core experience.

Using an open-weight model through local inference offers several advantages:

Greater control: Developers can choose and configure the model they want to run.
Privacy-conscious design: Prompts can be processed locally instead of being sent to a third-party AI provider, when the local setup is configured correctly.
Reduced dependence on external services: The application can use built-in missions even when AI is unavailable.
Learning and experimentation: Developers can inspect the workflow, experiment with prompts, and modify the application to suit their needs.
Accessibility for builders: A simple web application can serve as a starting point for experimenting with local AI.

A closed API could generate activity suggestions, too. However, local inference gives developers more control over where inference happens and how the AI component is configured.

For me, open innovation isn't only about making technology available. It's about giving people the freedom to understand it, experiment with it, and build something meaningful with it.

🤖 My Agent Session

I used AI-assisted development to help plan the application, develop its interface, and work through its implementation.

Agent session: Not published yet.

I plan to share a session link here if I create and publish one using DevRelay.

🏆 Prize Categories

Primary focus: Open-source AI application using local inference.

Potentially relevant partner categories: To be confirmed against the official challenge rules before submission.

🌍 Final Thoughts

TerraAI is my attempt to use AI for something that encourages people to disconnect from their devices and reconnect with the world around them.

This is an evolving project, and there is plenty of room to improve it—from more personalized missions and better activity tracking to a more robust local AI integration.

I'm excited to keep learning, building with open-source AI, and exploring how technology can encourage healthier digital habits.

Less scrolling. More exploring. More time outside. 🌿

Hacktoberfest #OpenSourceAI #TouchGrass #Ollama #Llama #WebDevelopment #BuildInPublic