DEV Community: Tejpunj Singh

Ghost in the Machine: Keeping Memories Alive with Open-Source AI

Tejpunj Singh — Tue, 27 Oct 2026 12:15:40 +0000

👻 Ghost in the Machine 🕯️

An immersive, open-source web experience built for Hacktoberfest 2026, capturing the eerie intersection of Halloween and Día de los Muertos (Day of the Dead).

This project explores the thin boundary between past and present—honoring forgotten digital histories, abandoned codebases, and loved ones through an interactive terminal and a glowing digital ofrenda.

🌐 Live Demo: https://tejpunjsingh.github.io/GHOST-IN-THE-MACHINE/

🎃 Core Features

1. 🖥️ The Haunted Terminal (Shell of the Past)

A custom command-line interface that "possesses" your browser session. When you interact with it, a simulated local AI model responds with spine-chilling, poetic, and mysterious commentary.

Interactive Commands: Type help, summon, status, hacktoberfest, about, or clear.
Atmospheric Logging: Color-coded terminal entries distinguishing between system states ([SYSTEM]), user inputs (guest@machine:~$), and ghostly machine whispers ([GHOST]).

2. 🕯️ Virtual Ofrenda Altar (Digital Remembrance)

Inspired by traditional Día de los Muertos altars, this feature allows you to build a personalized digital shrine right on your screen.

Custom Tributes: Input the name of a loved one, a historical figure, or even an obsolete programming language/tool you wish to honor.
Offerings & Memories: Add favorite foods, items, or memories alongside glowing CSS-animated digital candles and marigold flower borders (🌼 🕯️ 🌼).

🧠 Why Open Innovation Matters for This Project

In building Ghost in the Machine, an open-source and local-first approach was chosen for several critical reasons:

100% Offline Capability: The project runs entirely on your local machine inside the web browser. The machine spirit stays active and responsive even when the internet is completely disconnected—your device becomes the sanctuary.
Complete Data Privacy & Sovereignty: Personal stories, family memories, and heartfelt tributes entered into the virtual ofrenda remain strictly on your local device. No data is sent to a remote corporate server or stored in third-party clouds.
Freedom from Vendor Lock-In: Because it is built entirely using open web standards (HTML5, CSS3, and Vanilla JavaScript), it requires no paid APIs, subscription keys, or proprietary frameworks. It will never break due to a discontinued cloud service or changing corporate terms of service.
Extensibility & Hackability: Other developers can easily fork this repository, add their own custom ghost responses, modify the altar styles, or integrate actual local open-weight models via WebLLM or Ollama.

🛠️ Tech Stack

HTML5: Semantic document structure for accessibility and layout.
CSS3: Custom properties (:root), flexbox/grid layouts, and keyframe animations for flickering candle lighting effects and retro terminal aesthetics.
JavaScript (ES6+): Dynamic DOM manipulation, tab-switching controllers, event handling, and interactive command processing logic.

🚀 Getting Started & Local Setup

To run this project locally on your machine, follow these simple steps:

Clone the repository:


bash
   git clone [https://github.com/tejpunjsingh/GHOST-IN-THE-MACHINE.git](https://github.com/tejpunjsingh/GHOST-IN-THE-MACHINE.git)
![ ](https://dev-to-uploads.s3.us-east-2.amazonaws.com/uploads/articles/u8pjye9tk4u15rez7b6t.png)

Slow & Steady AI: Meet SlothAgent AI (LazyDesk) 🦥🚀

Tejpunj Singh — Thu, 22 Oct 2026 05:49:02 +0000

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

What I Built

Demo

2.amazonaws.com/uploads/articles/03akoi1bu7mngakop9zf.png)

Code

How I Built It

Why Does Open Innovation Matter?

My Agent Session

The Lazy Payoff

Prize Categories

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

What I Built

I built LazyDesk (SlothAgent AI) — a minimalist, privacy-first local AI workspace designed for intentional productivity.

Instead of overwhelming users with complex setups, LazyDesk offers a glassmorphism dashboard featuring focus timers, local task & note managers, and an offline-capable local AI assistant interface powered by Ollama. It helps developers, students, and creators organize their day and untangle thoughts without relying on external cloud subscriptions or sending personal data over the internet.

Demo

🌐 Live Website: https://tejpunjsingh.github.io/Lazy-AI-Assistant-/

Code

📁 GitHub Repository:

Tejpunjsingh / Lazy-AI-Assistant-

Lazy AI Assistant built with Open-Source AI for Hacktoberfest

LazyDesk AI

A local-first, low-friction workspace for the Slow & Steady challenge.
Tasks, notes, focus timer, and workspace backups work on your own computer. Optionally connect Ollama for open-weight AI assistance without a hosted AI API.

Features

Polished responsive dashboard with dark mode
Task list with categories, energy levels, completion, filtering and search
Quick notes saved locally in the browser
Focus timer with 5 / 15 / 25 minute presets
Optional AI chat powered by a model running in Ollama
Built-in rule-based fallback helpers when no model is installed
JSON export/import for your workspace
Offline app shell cache after first visit over localhost

Run it

Requirements: Python 3.9+ (no Python packages required).

Extract this folder.
Open a terminal in the lazydesk-ai folder.
Run:
```
python3 server.py
```
On Windows, use py server.py if python isn't available.
Open http://127.0.0.1:8000 in your browser.
Keep the terminal running while using the website.

No npm install…

View on GitHub

How I Built It

Frontend & UI: Single-file HTML, CSS (Tailwind & custom CSS for glassmorphism aesthetic), and Vanilla JS optimized for zero-configuration deployment on GitHub Pages.
Local AI Inference Engine: Integrated with Ollama running open-weight models locally (e.g., llama3.2).
Data Persistence: Uses browser localStorage for tasks, notes, and preferences—ensuring 100% client-side privacy.

Why Does Open Innovation Matter?

Open innovation and open-weight models make local-first computing truly accessible:

Total Data Privacy: Personal task dumps, daily notes, and prompts remain entirely on your device.
Zero API Cost: Users can leverage state-of-the-art open-source LLMs without monthly subscriptions or API key rate limits.
Offline Capability: Core task management and local AI inference function completely offline once downloaded.

The Lazy Payoff

LazyDesk eliminates context switching. Instead of opening 4 different apps for timer, tasks, notes, and AI chat, everything sits in a single, calm workspace. It saved me hours of daily planning and mental fatigue during Hacktoberfest project sprints!

Prize Categories

Overall Challenge
Open-Source AI / Local First Track

How I Built an AI-Powered Side Hustle Architect to Automate Startup Ideas 💡

Tejpunj Singh — Sat, 17 Oct 2026 08:23:45 +0000

bmission for the Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle*

What I Built

## What I Built

I built Open AI Side-Hustle Architect, a web-based tool that helps developers, students, and hackathon participants turn open-source AI ideas into practical project plans.

It helps users define problems, identify target customers, explore open-weight AI models and frameworks, estimate hardware requirements, evaluate offline capabilities, and generate structured Markdown hackathon submissions.

Demo

The project includes an interactive web interface for planning AI side hustles, evaluating project viability, and generating hackathon submissions.

Code

Built using HTML, Tailwind CSS, and JavaScript.

How I Built It

I developed the interface using HTML, Tailwind CSS, and JavaScript. The tool includes project presets, planning forms, a hardware estimation interface, and a Markdown submission generator.

It helps users explore project ideas involving open-weight AI models and local inference frameworks such as Llama and Ollama. The current version focuses on project planning and submission generation rather than running AI models locally.

Why Does Open Innovation Matter?

Open innovation gives developers more freedom to experiment with open-weight models and choose tools that fit their needs.

Depending on the model and hardware, local AI can help reduce recurring API costs, give users greater control over their data, and support offline workflows. This project encourages builders to consider these benefits while planning useful AI applications.

Demo

https://tejpunjsingh.github.io/Open-AI-Side-Hustle-Architect/

Code

How I Built It

Why Does Open Innovation Matter?

My Agent Session

My First User

Prize Categories

# 🌿 NatureBuddy AI — A Little More Outside

Tejpunj Singh — Sat, 10 Oct 2026 08:37:37 +0000

Live Website: https://tejpunjsingh.github.io/Nature-Buddy-AI-/

GitHub Repository: https://github.com/Tejpunjsingh/Nature-Buddy-AI-

💡 What inspired this project?

We spend so much time looking at screens that we sometimes forget to notice the world around us. A walk becomes another phone session, a garden becomes background scenery, and birds become sounds we barely hear.

I built NatureBuddy AI to turn that around. The idea is simple: use technology to encourage people to spend less time on their screens and more time exploring the natural world.

🌱 What does it do?

NatureBuddy is a nature companion that helps users turn spare moments into small outdoor adventures.

Personalized outdoor planning: Choose mindful walks, birdwatching, gardening, nature photography, cloud watching, or a relaxing outdoor break.
Flexible activities: Choose your setting, available time, and preferred mood to create a practical nature activity plan.
Discover nature: Search and filter ideas for outdoor activities based on your interests.
Private field notebook: Save observations, record small discoveries, delete notes, and export your journal.
Offline-friendly experience: Built-in activity guides work without a hosted AI API, and notes are stored locally in the browser.

The goal is to make the screen the shortest part of the experience: plan your adventure, step outside, and let the real world take over.

🧠 Why does open innovation matter?

Open innovation gives developers and users more control over the tools they build and how they use them.

NatureBuddy is designed around a local-first approach. Its activity guides run in the browser, while field notes stay on the user's device instead of requiring a cloud database. This reduces dependence on online services and helps protect personal observations.

My next step is to complete and test the integration of an open-weight model through Ollama. That would allow users to generate personalized plans locally, experiment with different models, and use AI without relying on a proprietary hosted inference service.

For this project, open innovation means making technology more transparent, adaptable, privacy-conscious, and accessible.

🛠️ Built with

HTML5
CSS3
JavaScript
Browser local storage
Open-weight AI integration planned with Ollama More noticing. A little more outside. 🌿

🌍 Why does it fit the Touch Grass theme?

NatureBuddy uses technology as a bridge to the outdoors rather than a destination in itself. It encourages simple, accessible activities that help people notice birds, plants, clouds, changing light, and the small details of everyday surroundings.

You don't need expensive equipment, a complicated schedule, or a perfect adventure. You just need a little curiosity.

🚀 What's next?

I plan to finish the local AI integration, improve offline support, expand the activity library, and test NatureBuddy during real outdoor walks. I also want to explore features such as locally generated nature quests and more detailed field journals.

My mission: Less scrolling.