Ghost in the Machine: AI Buddies as a Digital Ofrenda
This is a submission for the Hacktoberfest Open-Source AI Challenge Week 4: Ghost in the Machine
What I Built
For the final week of Hacktoberfest, combining the spirit of Día de los Muertos and Halloween, I built a feature extension inside AI Buddies that acts as a Virtual Ofrenda & Memory Guardian.
Instead of letting the code, contributions, and early lessons of open-source friends, mentors, and forgotten past projects fade away, this feature helps developers build a digital memorial and story archive. It channels open-weight models to act as a "ghost in the machine"—whispering back memories, preserving ancestral or early coding struggles, and telling stories of how past open-source tools paved the way for modern development.
Demo
- Live Platform: ai-buddies
- Live Memory Sandbox: Explore the virtual memory vault and interactive agent modules directly in your browser.
Code
{% github ajay-dhangar/ai-buddies %}
How I Built It
To capture that eerie, offline, reflective atmosphere, AI Buddies relies entirely on open-source AI foundations:
- Frontend Architecture: Built using React, TypeScript, Tailwind CSS, and custom styling to invoke a glowing, candle-lit digital ofrenda aesthetic with markdown and KaTeX support.
- Open AI & Agent Frameworks: Integrated local inference wrappers, LangChain agent harnesses, and Hugging Face open-weight models. This allows the memory-retrieval agent to run complex local narrative generation without hitting external servers.
- Offline Data Vault: Uses client-side storage paired with local model weights, ensuring personal stories, family oral histories, or remembrance notes never leave your personal machine.
Why Does Open Innovation Matter?
When dealing with personal memories, family archives, and tributes to loved ones, data privacy and control are non-negotiable:
1. Absolute Data Privacy: A closed commercial API would process family histories and personal memories on corporate servers over which we have no control. Open-weight models running locally or through trusted open agent harnesses keep all sentiment and personal records completely private.
2. Offline Permanence: Día de los Muertos honors what endures. Closed APIs can deprecate endpoints, change terms of service, or shut down overnight. An open-source approach means this memorial tool can run on a laptop with zero internet connection, preserving memories indefinitely.
3. Custom "Spooky" Tuning: Closed chat models enforce strict neutral corporate personas. Open-weight models let us rewrite system prompts completely, allowing the agent to channel historical developer lore and respectful remembrance tones authentically.
My Agent Session
(Optional: Insert your saved DevRelay session link or tag here if you recorded your development workflow!)
The Spirit of It
When I tested the virtual memory ofrenda with a fellow developer by feeding in notes about an old, abandoned open-source project we used to work on years ago, the local model generated a hauntingly accurate recap of our old debugging nights, complete with nostalgic styling. It felt like bringing a digital ghost back to life—reminding us that behind every repository lies human effort, memory, and community.
Prize Categories
- Main Track: Ghost in the Machine
- Open Source AI / Local Inference Track
AI Buddies: Lazy-Mode Learning for Developers
This is a submission for the Hacktoberfest Open-Source AI Challenge Week 3: Slow and Steady
What I Built
Learning complex algorithms and navigating huge codebases takes hours of tedious reading, manually tracing variable changes on paper, and digging through endless stack traces.
I built AI Buddies to make learning and debugging entirely effortless. Instead of forcing you to read walls of text or debug step-by-step manually, AI Buddies automates the entire comprehension workflow. It visualizes data structures dynamically, breaks down complex logic into bite-sized summaries, and guides you through problem-solving without making you scratch your head for hours.
Who gets to be a little lazier because of it?
- Computer Science Students: Skip spending 3 hours tracing recursion trees on paper—AI Buddies renders the execution tree instantly.
- Self-Taught Developers: Stop opening 20 browser tabs to understand a single algorithm concept—get step-by-step hints served directly in your sandbox.
- Open-Source Maintainers: Automate onboarding and documentation breakdowns so new contributors don't constantly ask basic setup questions.
Demo
- Live Platform: ai-buddies
- Interactive Visualizers: Trace algorithms and test interactive agentic roadmaps effortlessly in your browser.
Code
{% github ajay-dhangar/ai-buddies %}
How I Built It
To keep response times lightning fast and fully automated, AI Buddies relies on a modern open-source AI architecture:
- Frontend & Visual Engine: Built with React, TypeScript, Tailwind CSS, and Docusaurus. Custom visual components render interactive step-by-step flowcharts and KaTeX math formulas automatically.
- Open AI & Agent Harnesses: Powered by open-weight model integration and agent frameworks (such as Hugging Face open weights, LangChain agent pipelines, and local inference via Ollama).
- Automated Routing: Agents parse user queries or code snippets, auto-select the best open model for the job (fast syntax breakdown vs. full logic tracing), and return interactive UI previews instead of long, boring text responses.
Why Does Open Innovation Matter?
"Slow and Steady" means building sustainable tools that run anywhere, anytime, without relying on unpredictable third-party services or recurring API bills. Open innovation makes this possible:
1. Zero Operating Overhead: Closed APIs charge per token and per request, which punishes users who learn at their own pace. Open weights and local inference let students run agentic workflows continuously for free.
2. Offline & Low-Spec Execution: Lightweight open models run directly on modest hardware without needing high-speed internet. You can learn on a train, in a low-bandwidth area, or on an older laptop.
3. Custom Agent Rules: Closed models often try to solve the entire problem instantly, taking away the learning opportunity. Open frameworks let us explicitly program agentic behaviors to gently guide, hint, and visualize.
My Agent Session
(Optional: Insert your saved DevRelay session link or tag here if you recorded your development workflow!)
The Lazy Payoff
This week, I used AI Buddies to walk through a complex graph traversal problem that usually takes 45–60 minutes of manual diagramming and trial-and-error debugging.
- Time Saved: Cut understanding and implementation time down from 1 hour to under 10 minutes.
- Effort Saved: Zero paper diagrams needed. The automated visualizer rendered node visit sequences and queue states step-by-step with a single click.
Prize Categories
- Main Track: Slow and Steady
- Open Source AI / Local Inference Track
Building AI Buddies: An Open AI Learning Companion for CodeHarborHub
_This is a submission for the MLH x DEV Writing Challenge_
What I Built
Learning Data Structures, Algorithms, and full-stack development can feel daunting when you're navigating complex documentation alone. As the maintainer of CodeHarborHub, an open-source platform dedicated to empowering developers through structured learning paths. I wanted to create an interactive tool to make self-paced technical education truly engaging.
I built AI Buddies, an open-source AI learning partner integrated within the CodeHarborHub ecosystem. Instead of acting as a typical chatbot that dumps ready-made answers, AI Buddies provides step-by-step guidance, breaks down algorithm concepts with real-time visualizers, and acts as a patient partner for learners solving coding challenges in an interactive sandbox.
Key Learnings:
- Designing non-intrusive AI agents that guide learning rather than completing tasks for the user.
- Structuring lightweight MDX and React execution environments for seamless browser rendering.
- Balancing client-side interactivity with open-weight model architectures.
Demo
- Live Platform: AI Buddies
- Main Ecosystem: CodeHarborHub
- Interactive Roadmaps: Explore live data structure visualizers and agentic learning modules directly in the browser.
Code
{% github ajay-dhangar/ai-buddies %}
You can also check out our wider open-source organization:
{% github CodeHarborHub/codeharborhub %}
Partner Technologies
To keep the platform fast, responsive, and open for global contributors, I utilized several core technologies and developer platforms:
- GitHub & GitHub Actions: Used for hosting our repositories, managing open-source contributor pull requests, and orchestrating CI/CD pipelines to build and deploy docs and interactive components automatically.
- Open AI & Agent Harnesses: Integrated open-weight model routing and agent framework concepts (such as LangChain and Hugging Face open weights), enabling learners to leverage model flexibility without vendor lock-in.
- Frontend Stack: Built using React, TypeScript, Tailwind CSS, and Docusaurus for fast, modern documentation layouts with interactive live code previews and KaTeX notation support.
Implementing these tools allowed us to maintain zero-cost inference paths for students while giving them full control over model customization and data privacy.
Hackathon Experience
Participating in hackathons as part of the MLH community is always an invigorating experience. The collaborative atmosphere, late-night debugging sessions, and feedback loops with fellow hackers inspired many of the features now powering AI Buddies.
What I’ll remember most is sharing our live visualizers with student developers during community review rounds. Seeing learners instantly grasp a complex tree traversal or dynamic programming concept through an interactive agent session reminded me why open-source community building matters so much.
---
If you're interested in open-source AI or building developer tools, check out our repository and drop a star or PR!
AI Buddies: An Open AI Learning Micro-SaaS for Developers
This is a submission for the Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle
What I Built
Self-taught developers, computer science students, and bootcamp grads spend hundreds of dollars on private tutors or generic AI chat subscriptions that just output pre-written code without explaining how or why it works.
I built AI Buddies as a Freemium Micro-SaaS learning partner platform. It offers guided, step-by-step Data Structures, Algorithms, and full-stack interactive paths powered by open-source AI.
Why would people pay for it?
While basic AI chatbots give flat answers, AI Buddies acts as an interactive tutor with real-time code visualizers, sandbox exercises, and step-by-step hint systems. Students and job seekers pay a small monthly subscription ($5–$10/month) for premium feature modules—such as personalized mock technical interview simulations, custom repository code analysis, and unlimited agentic learning workflows—at a fraction of the cost of traditional tutoring platforms.
Demo
- Live Platform: ai-buddies
- Interactive Visualizers: Try interactive roadmaps and agentic study paths live in your browser.
Code
{% github ajay-dhangar/ai-buddies %}
How I Built It
To keep operational overhead low enough for a profitable side hustle, AI Buddies relies on an open-source AI stack:
- Frontend & Web UI: Built with React, TypeScript, Tailwind CSS, and Docusaurus for fast, lightweight rendering and interactive documentation layout.
- Open AI & Agent Frameworks: Integrated open-weight model routing (via Hugging Face and local inference runners like Ollama / LangChain agent harnesses). This allows learners to connect their own local hardware or access lightweight hosted open weights without expensive API middleware.
- Interactive Learning Tools: Built custom React and KaTeX components for mathematical notation, real-time code visualization, and algorithmic execution tracking.
Why Does Open Innovation Matter?
In a side hustle, unit economics make or break the business. Open innovation gave AI Buddies three distinct advantages over proprietary closed APIs:
1. Zero-Inference Cost Base: Proprietary LLM APIs charge per token, making high-volume student usage expensive. Using open weights and client-side or local inference options keeps server costs near zero, making a freemium pricing tier sustainable.
2. Data Privacy & Local Running: Learners often work with proprietary interview questions or sensitive codebases. Open-weight models run locally on the user's machine, keeping their data off third-party servers.
3. No Vendor Lock-In: Open agent harnesses let us swap specialized open-weight models (e.g., code-tuned weights vs. general explanation models) instantly without rewriting prompt pipelines or accepting price hikes.
My Agent Session
(Optional: Insert your saved DevRelay session link or tag here if you recorded your development workflow!)
My First User
I shared the live AI Buddies platform with a developer from my community who was preparing for technical interviews.
What they said:
> "Regular AI chats just give away the solution immediately, which ruins the learning process for DSA. AI Buddies actually walks me through debugging step-by-step with visual diagrams. I would easily pay $10/month for the interview simulation module alone because it's cheaper than a single hour with a human tutor."
Prize Categories
- Main Track: Side Hustle
- Best Use of Gemma (If deploying Gemma open weights)
- Best Use of DigitalOcean / Render (Depending on your hosting provider)
Mentions
Ghost in the Machine: AI Buddies as a Digital Ofrenda This is a submission for the Hacktoberfest Open-Source AI Challenge Week 4: Ghost in the Machine What I Built For the final week of Hacktoberfest, combining the spirit of Día de los Muertos and Halloween, I built a feature extension inside AI Buddies that acts as a Virtual Ofrenda & Memory Guardian. Instead of letting the code, contributions, and early lessons of open-source friends, mentors, and forgotten past projects fade away, this feature helps developers build a digital memorial and story archive. It channels open-weight models to act as a "ghost in the machine"—whispering back memories, preserving ancestral or early coding struggles, and telling stories of how past open-source tools paved the way for modern development. Demo - Live Platform: ai-buddies - Live Memory Sandbox: Explore the virtual memory vault and interactive agent modules directly in your browser. Code {% github ajay-dhangar/ai-buddies %} How I Built It To capture that eerie, offline, reflective atmosphere, AI Buddies relies entirely on open-source AI foundations: - Frontend Architecture: Built using React, TypeScript, Tailwind CSS, and custom styling to invoke a glowing, candle-lit digital ofrenda aesthetic with markdown and KaTeX support. - Open AI & Agent Frameworks: Integrated local inference wrappers, LangChain agent harnesses, and Hugging Face open-weight models. This allows the memory-retrieval agent to run complex local narrative generation without hitting external servers. - Offline Data Vault: Uses client-side storage paired with local model weights, ensuring personal stories, family oral histories, or remembrance notes never leave your personal machine. Why Does Open Innovation Matter? When dealing with personal memories, family archives, and tributes to loved ones, data privacy and control are non-negotiable: 1. Absolute Data Privacy: A closed commercial API would process family histories and personal memories on corporate servers over which we have no control. Open-weight models running locally or through trusted open agent harnesses keep all sentiment and personal records completely private. 2. Offline Permanence: Día de los Muertos honors what endures. Closed APIs can deprecate endpoints, change terms of service, or shut down overnight. An open-source approach means this memorial tool can run on a laptop with zero internet connection, preserving memories indefinitely. 3. Custom "Spooky" Tuning: Closed chat models enforce strict neutral corporate personas. Open-weight models let us rewrite system prompts completely, allowing the agent to channel historical developer lore and respectful remembrance tones authentically. My Agent Session (Optional: Insert your saved DevRelay session link or tag here if you recorded your development workflow!) The Spirit of It When I tested the virtual memory ofrenda with a fellow developer by feeding in notes about an old, abandoned open-source project we used to work on years ago, the local model generated a hauntingly accurate recap of our old debugging nights, complete with nostalgic styling. It felt like bringing a digital ghost back to life—reminding us that behind every repository lies human effort, memory, and community. Prize Categories - Main Track: Ghost in the Machine - Open Source AI / Local Inference Track
AI Buddies: Lazy-Mode Learning for Developers This is a submission for the Hacktoberfest Open-Source AI Challenge Week 3: Slow and Steady What I Built Learning complex algorithms and navigating huge codebases takes hours of tedious reading, manually tracing variable changes on paper, and digging through endless stack traces. I built AI Buddies to make learning and debugging entirely effortless. Instead of forcing you to read walls of text or debug step-by-step manually, AI Buddies automates the entire comprehension workflow. It visualizes data structures dynamically, breaks down complex logic into bite-sized summaries, and guides you through problem-solving without making you scratch your head for hours. Who gets to be a little lazier because of it? - Computer Science Students: Skip spending 3 hours tracing recursion trees on paper—AI Buddies renders the execution tree instantly. - Self-Taught Developers: Stop opening 20 browser tabs to understand a single algorithm concept—get step-by-step hints served directly in your sandbox. - Open-Source Maintainers: Automate onboarding and documentation breakdowns so new contributors don't constantly ask basic setup questions. Demo - Live Platform: ai-buddies - Interactive Visualizers: Trace algorithms and test interactive agentic roadmaps effortlessly in your browser. Code {% github ajay-dhangar/ai-buddies %} How I Built It To keep response times lightning fast and fully automated, AI Buddies relies on a modern open-source AI architecture: - Frontend & Visual Engine: Built with React, TypeScript, Tailwind CSS, and Docusaurus. Custom visual components render interactive step-by-step flowcharts and KaTeX math formulas automatically. - Open AI & Agent Harnesses: Powered by open-weight model integration and agent frameworks (such as Hugging Face open weights, LangChain agent pipelines, and local inference via Ollama). - Automated Routing: Agents parse user queries or code snippets, auto-select the best open model for the job (fast syntax breakdown vs. full logic tracing), and return interactive UI previews instead of long, boring text responses. Why Does Open Innovation Matter? "Slow and Steady" means building sustainable tools that run anywhere, anytime, without relying on unpredictable third-party services or recurring API bills. Open innovation makes this possible: 1. Zero Operating Overhead: Closed APIs charge per token and per request, which punishes users who learn at their own pace. Open weights and local inference let students run agentic workflows continuously for free. 2. Offline & Low-Spec Execution: Lightweight open models run directly on modest hardware without needing high-speed internet. You can learn on a train, in a low-bandwidth area, or on an older laptop. 3. Custom Agent Rules: Closed models often try to solve the entire problem instantly, taking away the learning opportunity. Open frameworks let us explicitly program agentic behaviors to gently guide, hint, and visualize. My Agent Session (Optional: Insert your saved DevRelay session link or tag here if you recorded your development workflow!) The Lazy Payoff This week, I used AI Buddies to walk through a complex graph traversal problem that usually takes 45–60 minutes of manual diagramming and trial-and-error debugging. - Time Saved: Cut understanding and implementation time down from 1 hour to under 10 minutes. - Effort Saved: Zero paper diagrams needed. The automated visualizer rendered node visit sequences and queue states step-by-step with a single click. Prize Categories - Main Track: Slow and Steady - Open Source AI / Local Inference Track
Building AI Buddies: An Open AI Learning Companion for CodeHarborHub _This is a submission for the MLH x DEV Writing Challenge_ What I Built Learning Data Structures, Algorithms, and full-stack development can feel daunting when you're navigating complex documentation alone. As the maintainer of CodeHarborHub, an open-source platform dedicated to empowering developers through structured learning paths. I wanted to create an interactive tool to make self-paced technical education truly engaging. I built AI Buddies, an open-source AI learning partner integrated within the CodeHarborHub ecosystem. Instead of acting as a typical chatbot that dumps ready-made answers, AI Buddies provides step-by-step guidance, breaks down algorithm concepts with real-time visualizers, and acts as a patient partner for learners solving coding challenges in an interactive sandbox. Key Learnings: - Designing non-intrusive AI agents that guide learning rather than completing tasks for the user. - Structuring lightweight MDX and React execution environments for seamless browser rendering. - Balancing client-side interactivity with open-weight model architectures. Demo - Live Platform: AI Buddies - Main Ecosystem: CodeHarborHub - Interactive Roadmaps: Explore live data structure visualizers and agentic learning modules directly in the browser. Code {% github ajay-dhangar/ai-buddies %} You can also check out our wider open-source organization: {% github CodeHarborHub/codeharborhub %} Partner Technologies To keep the platform fast, responsive, and open for global contributors, I utilized several core technologies and developer platforms: - GitHub & GitHub Actions: Used for hosting our repositories, managing open-source contributor pull requests, and orchestrating CI/CD pipelines to build and deploy docs and interactive components automatically. - Open AI & Agent Harnesses: Integrated open-weight model routing and agent framework concepts (such as LangChain and Hugging Face open weights), enabling learners to leverage model flexibility without vendor lock-in. - Frontend Stack: Built using React, TypeScript, Tailwind CSS, and Docusaurus for fast, modern documentation layouts with interactive live code previews and KaTeX notation support. Implementing these tools allowed us to maintain zero-cost inference paths for students while giving them full control over model customization and data privacy. Hackathon Experience Participating in hackathons as part of the MLH community is always an invigorating experience. The collaborative atmosphere, late-night debugging sessions, and feedback loops with fellow hackers inspired many of the features now powering AI Buddies. What I’ll remember most is sharing our live visualizers with student developers during community review rounds. Seeing learners instantly grasp a complex tree traversal or dynamic programming concept through an interactive agent session reminded me why open-source community building matters so much. --- If you're interested in open-source AI or building developer tools, check out our repository and drop a star or PR!
AI Buddies: An Open AI Learning Micro-SaaS for Developers This is a submission for the Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle What I Built Self-taught developers, computer science students, and bootcamp grads spend hundreds of dollars on private tutors or generic AI chat subscriptions that just output pre-written code without explaining how or why it works. I built AI Buddies as a Freemium Micro-SaaS learning partner platform. It offers guided, step-by-step Data Structures, Algorithms, and full-stack interactive paths powered by open-source AI. Why would people pay for it? While basic AI chatbots give flat answers, AI Buddies acts as an interactive tutor with real-time code visualizers, sandbox exercises, and step-by-step hint systems. Students and job seekers pay a small monthly subscription ($5–$10/month) for premium feature modules—such as personalized mock technical interview simulations, custom repository code analysis, and unlimited agentic learning workflows—at a fraction of the cost of traditional tutoring platforms. Demo - Live Platform: ai-buddies - Interactive Visualizers: Try interactive roadmaps and agentic study paths live in your browser. Code {% github ajay-dhangar/ai-buddies %} How I Built It To keep operational overhead low enough for a profitable side hustle, AI Buddies relies on an open-source AI stack: - Frontend & Web UI: Built with React, TypeScript, Tailwind CSS, and Docusaurus for fast, lightweight rendering and interactive documentation layout. - Open AI & Agent Frameworks: Integrated open-weight model routing (via Hugging Face and local inference runners like Ollama / LangChain agent harnesses). This allows learners to connect their own local hardware or access lightweight hosted open weights without expensive API middleware. - Interactive Learning Tools: Built custom React and KaTeX components for mathematical notation, real-time code visualization, and algorithmic execution tracking. Why Does Open Innovation Matter? In a side hustle, unit economics make or break the business. Open innovation gave AI Buddies three distinct advantages over proprietary closed APIs: 1. Zero-Inference Cost Base: Proprietary LLM APIs charge per token, making high-volume student usage expensive. Using open weights and client-side or local inference options keeps server costs near zero, making a freemium pricing tier sustainable. 2. Data Privacy & Local Running: Learners often work with proprietary interview questions or sensitive codebases. Open-weight models run locally on the user's machine, keeping their data off third-party servers. 3. No Vendor Lock-In: Open agent harnesses let us swap specialized open-weight models (e.g., code-tuned weights vs. general explanation models) instantly without rewriting prompt pipelines or accepting price hikes. My Agent Session (Optional: Insert your saved DevRelay session link or tag here if you recorded your development workflow!) My First User I shared the live AI Buddies platform with a developer from my community who was preparing for technical interviews. What they said: > "Regular AI chats just give away the solution immediately, which ruins the learning process for DSA. AI Buddies actually walks me through debugging step-by-step with visual diagrams. I would easily pay $10/month for the interview simulation module alone because it's cheaper than a single hour with a human tutor." Prize Categories - Main Track: Side Hustle - Best Use of Gemma (If deploying Gemma open weights) - Best Use of DigitalOcean / Render (Depending on your hosting provider)