DEV Community: Vaibhav Yadav

Digital Afterlife — Reviving Abandoned Code with Open-Source AI

Vaibhav Yadav — Tue, 27 Oct 2026 08:44:33 +0000

👻 Digital Afterlife: Reviving Abandoned Code with Open-Source AI

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

What I Built

Every abandoned codebase has a story. A project that once held someone's ideas, late-night coding sessions, and big ambitions can eventually become a forgotten folder on a computer.

Digital Afterlife is a ghost-themed prototype designed to help developers bring forgotten projects back to life.

Inspired by the Ghost in the Machine theme, I built a tool that scans a selected project for clues about its past and identifies opportunities to revive it.

The prototype can:

👻 Scan source files for TODO, FIXME, HACK, and XXX markers.
🔍 Look for basic project signals, such as README files, tests, licenses, and .gitignore.
📊 Generate a simple project-health score.
🧭 Suggest practical next steps for bringing a project back to life.
💻 Analyze files selected in the browser without uploading them to a server.

The idea is simple: what if AI could help developers rediscover the potential hidden in their forgotten code?

The current version uses local file analysis and rule-based logic. Integrating an open-weight AI model for deeper code understanding is the next step.

Demo

🚧 Live demo: file:///D:/College%20Studies/Self%20Work/Hacktober/index.html

A short screen-recorded demonstration would also help showcase the file scanning, findings, and resurrection plan.

Code

💻 GitHub Repository: https://github.com/vaibhavyadavoefh-pixel/Digital-Afterlife-Reviving-Abandoned-Code-with-Open-Source-AI

The project currently uses:

HTML for the interface
CSS for the visual design and animations
JavaScript for file analysis and recommendations

The codebase is designed to evolve toward a backend powered by Python, FastAPI, and an open-weight AI model.

How I Built It

I built Digital Afterlife using HTML, CSS, and JavaScript, focusing on creating a functional prototype with a dark, ghost-inspired interface.

The development process involved:

Designing a dashboard that makes code analysis easy to understand.
Implementing browser-based file selection and source-code scanning.
Detecting common development markers and checking for basic project files.
Creating a project-health score and a list of suggested recovery steps.
Adding a demo mode to demonstrate the experience without requiring an existing project.

AI integration is still planned, not implemented in this version.

My next step is to integrate an open-weight model that can inspect code context, explain unfinished features, identify potential bugs, and recommend meaningful improvements. I want to explore local inference so developers can analyze their projects without depending entirely on paid, closed AI APIs.

Why Does Open Innovation Matter?

Open innovation makes AI more accessible to developers who are still learning and experimenting.

As a first-year B.Tech CSE student specializing in AI and ML, I'm exploring how open-source tools can help turn ideas into practical projects.

For Digital Afterlife, open-source AI could make it possible to:

Run code analysis using models developers can inspect and experiment with.
Customize the AI for different programming languages and project types.
Explore local inference and greater control over source-code privacy.
Allow contributors to improve the tool, add integrations, and build new features.
Make AI-assisted project recovery more accessible without requiring every developer to depend on a closed API.

A closed API might help build a similar tool, but an open approach gives developers more freedom to experiment with the model, customize the workflow, and contribute improvements to the community.

I want Digital Afterlife to become more than a code scanner. I want it to become an open-source companion that helps developers recover unfinished ideas and continue where they left off.

My Agent Session

I used AI-assisted development while building the prototype.

DevRelay session: [Add your saved agent session link here, if available]

I plan to document the next development phase, including the integration of an open-weight model and the reasoning behind the implementation choices.

The Spirit of It 👻

Every developer has probably started a project they were excited about, only to leave it unfinished when another idea came along.

Sometimes, the code is still there. The original motivation isn't.

Digital Afterlife is built around that little moment of rediscovery: opening an old project, finding an unfinished feature, and realizing that the idea might still be worth pursuing.

Instead of treating abandoned code as a graveyard, what if we treated it as a place full of unfinished possibilities?

The code isn't dead. It might just be waiting for another commit. 💚

Prize Categories

Update this section according to the challenge's official partner categories and your project's actual eligibility.

Open-source AI / local AI category, if applicable.
Any other eligible partner category supported by the final implementation.

Built with curiosity, a little Halloween spirit, and the belief that forgotten projects deserve another chance. 👻

OpenStudy AI — Study Smarter. Do Less.

Vaibhav Yadav — Tue, 20 Oct 2026 11:11:20 +0000

✦ OpenStudy AI — Study Smarter. Do Less.

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

What I Built

Meet OpenStudy AI, a personal AI study assistant designed to help students spend less time organizing notes and more time learning.

As students, we spend hours turning lengthy lecture notes into summaries, extracting important concepts, and preparing revision questions. OpenStudy AI aims to automate this repetitive work using locally running AI.

With OpenStudy AI, students can:

📝 Summarize Notes: Transform lengthy study material into concise revision notes.
💡 Extract Key Points: Identify important concepts and definitions.
🧠 Generate Revision Questions: Create practice questions and answers from study material.
📚 Simplify Complex Topics: Get beginner-friendly explanations of difficult concepts.

The idea is simple: let AI handle the repetitive work so students can focus on understanding what they study.

Demo

Live Demo: https://vaibhavyadavoefh-pixel.github.io/OpenStudy-AI/

Screenshots:

The application requires a running local Ollama instance and a downloaded model to generate AI responses.

Code

GitHub Repository: https://github.com/vaibhavyadavoefh-pixel/OpenStudy-AI.git

The project uses HTML, CSS, and JavaScript for its interface and application logic, with Ollama providing access to a locally running AI model.

Contributions, improvements, and new ideas are welcome!

How I Built It

I built OpenStudy AI using:

HTML5 — Application structure.
CSS3 — Responsive interface with a modern dark theme.
JavaScript — Application logic and communication with the AI model.
Ollama — Running an AI model locally.
Qwen2.5 3B — The planned default open-weight model for generating study material.

The workflow is straightforward: students paste their notes, choose a task, and receive AI-generated study material.

Instead of sending every request to a paid cloud AI API, the application communicates with a model running on the user's machine.

This makes local inference the core of the project rather than just adding an AI label to a regular productivity app.

Why Does Open Innovation Matter?

I believe useful AI tools should be accessible to students without requiring them to purchase API subscriptions or automatically send their study material to external AI services.

Using open-weight models and local inference offers several advantages:

Privacy and control: Study material can be processed on the user's own machine.
Lower running costs: No mandatory paid cloud AI API subscription is required.
Flexibility: Users can experiment with different compatible local models.
Customization: Prompts and application behavior can be modified to suit different learning styles.
Offline potential: Once the model and required application resources are downloaded, AI inference can work without an internet connection.

Closed AI services can offer excellent performance, but they may introduce usage limits, recurring costs, or dependence on an external provider.

OpenStudy AI explores an alternative: an adaptable study assistant that students can run and experiment with themselves.

Open innovation makes this approach possible by giving developers more control over the models and tools they build with.

My Agent Session

DevRelay Session: [Add your saved DevRelay agent session link here]

I used AI-assisted development to help build the project and explore how open-source AI can solve an everyday student problem.

The goal was not simply to build another chatbot, but to create something useful around a specific task that students repeatedly perform.

The Lazy Payoff

Imagine having a long set of lecture notes the night before a test.

Instead of manually creating a summary, extracting definitions, and writing revision questions, you can give the repetitive parts to OpenStudy AI.

The potential payoff is less time spent preparing revision material and more time available for actually studying it.

I haven't measured the time saved yet, so this is the intended benefit rather than a verified result. Testing it with real coursework will help me understand how much effort it can genuinely save.

Prize Categories

Primary category: Open-Source AI

OpenStudy AI uses an open-weight model through local inference, making the AI component central to the application's functionality.

I'll update this section with any additional applicable partner categories after verifying the challenge requirements.

Final Thoughts

OpenStudy AI is my attempt to make studying a little less repetitive and a little more accessible.

This project is also an opportunity for me to learn by building something practical with open-source AI.

Sometimes, working smarter means knowing which tasks you don't need to do manually.

Slow and steady. Let AI handle the boring stuff. ✦

Campus Grass Challenge

Vaibhav Yadav — Thu, 08 Oct 2026 09:51:45 +0000

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

Demo

🌱 Live Demo: https://vaibhavyadavoefh-pixel.github.io/TOUCH-GRASS/

The prototype is fully browser-based and requires no account or backend.

Code

💻 GitHub Repository: https://github.com/vaibhavyadavoefh-pixel/TOUCH-GRASS

The project is built with simple web technologies so that it remains lightweight, understandable, and easy to modify.

How I Built It

The current prototype is built using:

HTML
CSS
Vanilla JavaScript
GitHub Pages

The project intentionally avoids unnecessary frameworks and backend services.

The current prototype focuses on the core Touch Grass experience: daily missions, progress tracking, XP, streaks, screen-time logging, and a leaderboard.

The next stage is to integrate a local open-weight AI model so the challenges can become personalized instead of being a fixed list.

For example, a local AI could generate challenges based on:

Available free time
Previous completed challenges
Campus environment
User preferences
Difficulty level
Previous activity

The goal is to eventually run this intelligence locally rather than depending on a closed cloud API.

Why Does Open Innovation Matter?

The idea behind Campus Grass is somewhat ironic: the app should ultimately help you use your screen less.

Using open-source AI makes it possible to move the intelligence closer to the user instead of sending personal activity information to a closed third-party service.

A local/open-weight approach can provide:

Better privacy for personal activity data
Offline functionality
Freedom to change or experiment with models
No dependency on a single AI provider
The ability to fine-tune or adapt the system for different campuses
Lower operating costs for an open-source project

For a project built around getting people away from screens, having an AI system that can eventually run locally is especially interesting.

My Agent Session

Not included in the current submission.

Prize Categories

Not entering any additional partner categories at this time.

The Idea

The goal isn't to use this app more.
The goal is to use it less.

Campus Grass turns real-world activities into small challenges and gives students a reason to step outside, explore their campus, interact with people, and spend intentional time away from their screens.

Built for the Hacktoberfest Open-Source AI Challenge — Touch Grass.