<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: Parth Ingle</title>
    <description>The latest articles on DEV Community by Parth Ingle (@parth_ingle).</description>
    <link>https://dev.to/parth_ingle</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4149873%2Fe3020d5f-fcb0-4126-bbce-980ae59d7483.png</url>
      <title>DEV Community: Parth Ingle</title>
      <link>https://dev.to/parth_ingle</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/parth_ingle"/>
    <language>en</language>
    <item>
      <title>👻 I Built Bug Exorcist: An AI Debugging Agent That Exorcises Bugs Locally with Gemma</title>
      <dc:creator>Parth Ingle</dc:creator>
      <pubDate>Sat, 31 Oct 2026 18:12:04 +0000</pubDate>
      <link>https://dev.to/parth_ingle/i-built-bug-exorcist-an-ai-debugging-agent-that-exorcises-bugs-locally-with-gemma-1203</link>
      <guid>https://dev.to/parth_ingle/i-built-bug-exorcist-an-ai-debugging-agent-that-exorcises-bugs-locally-with-gemma-1203</guid>
      <description>&lt;p&gt;&lt;em&gt;This is a submission for the &lt;a href="https://dev.to/challenges/hacktoberfest-week4-2026-10-26"&gt;Hacktoberfest Open-Source AI Challenge Week 4: Ghost in the Machine&lt;/a&gt;&lt;/em&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  What I Built
&lt;/h2&gt;

&lt;p&gt;👻 Bug Exorcist — Cast out bugs. Bring your code back to life.&lt;/p&gt;

&lt;p&gt;We've all experienced it: you're working on a project, everything seems fine, and suddenly your terminal throws an error that makes absolutely no sense.&lt;/p&gt;

&lt;p&gt;You inspect multiple files, read stack traces, try different solutions, and somehow end up with even more bugs.&lt;/p&gt;

&lt;p&gt;That's what inspired me to build Bug Exorcist, a local-first AI debugging assistant powered by Google's open-weight Gemma model.&lt;/p&gt;

&lt;p&gt;Instead of simply explaining an error message, Bug Exorcist aims to investigate the actual project, identify relevant source files, understand the error context, and suggest possible fixes.&lt;/p&gt;

&lt;p&gt;Here's how it works:&lt;br&gt;
👻 Summon Your Project: Upload a project ZIP file and let Bug Exorcist inspect its structure.&lt;/p&gt;

&lt;p&gt;🔍 Identify the Possession: Submit an error message, traceback, or stack trace.&lt;/p&gt;

&lt;p&gt;🧠 Investigate the Bug: Gemma analyzes the relevant code and debugging context.&lt;/p&gt;

&lt;p&gt;🪄 Perform the Exorcism: Receive a root-cause explanation and suggested code correction.&lt;/p&gt;

&lt;p&gt;The initial version focuses on Python, JavaScript, and TypeScript projects.&lt;/p&gt;

&lt;p&gt;The idea is simple: developers shouldn't just receive an explanation of an error. They should be able to understand where it originated and how to fix it.&lt;/p&gt;

&lt;h2&gt;
  
  
  Demo
&lt;/h2&gt;

&lt;p&gt;Bug Exorcist is designed as a locally running application rather than a hosted AI service.&lt;/p&gt;

&lt;h2&gt;
  
  
  Code
&lt;/h2&gt;

&lt;p&gt;👨‍💻 GitHub Repository: &lt;a href="https://github.com/parthingle-codes/Bug-Exorcist" rel="noopener noreferrer"&gt;GITHUB REPOSITORY&lt;/a&gt;&lt;br&gt;
The project uses a React frontend and a Python FastAPI backend, with Ollama providing local access to Gemma.&lt;/p&gt;

&lt;p&gt;Contributions, feedback, and suggestions are welcome!&lt;/p&gt;

&lt;h2&gt;
  
  
  How I Built It
&lt;/h2&gt;

&lt;p&gt;The central part of Bug Exorcist is Gemma running locally through Ollama.&lt;/p&gt;

&lt;p&gt;I wanted to build something where open-weight AI wasn't simply an additional feature but the technology that made the application possible.&lt;/p&gt;

&lt;p&gt;Here's the technology stack:&lt;/p&gt;

&lt;p&gt;Component&lt;/p&gt;

&lt;p&gt;Technology&lt;/p&gt;

&lt;p&gt;Frontend&lt;/p&gt;

&lt;p&gt;React + Vite&lt;/p&gt;

&lt;p&gt;Backend&lt;/p&gt;

&lt;p&gt;Python + FastAPI&lt;/p&gt;

&lt;p&gt;AI Model&lt;/p&gt;

&lt;p&gt;Google Gemma 3 1B&lt;/p&gt;

&lt;p&gt;Model Runtime&lt;/p&gt;

&lt;p&gt;Ollama&lt;/p&gt;

&lt;p&gt;Storage&lt;/p&gt;

&lt;p&gt;Local project files&lt;/p&gt;

&lt;p&gt;The debugging workflow follows a straightforward pipeline:&lt;/p&gt;

&lt;p&gt;The developer uploads a project ZIP file.&lt;/p&gt;

&lt;p&gt;The backend extracts and inspects the project structure.&lt;/p&gt;

&lt;p&gt;The application detects supported programming languages and processes the submitted error.&lt;/p&gt;

&lt;p&gt;Relevant source-code context is collected.&lt;/p&gt;

&lt;p&gt;The context is sent to Gemma through the local Ollama runtime.&lt;/p&gt;

&lt;p&gt;Gemma generates a structured diagnosis with a likely root cause and suggested correction.&lt;/p&gt;

&lt;p&gt;The frontend displays the investigation results.&lt;/p&gt;

&lt;p&gt;I'm also working towards adding a proper evidence board, code-diff preview, patch approval, and automated verification.&lt;/p&gt;

&lt;p&gt;The intention is to make Bug Exorcist an actual debugging companion rather than another chatbot with a fancy interface.&lt;/p&gt;

&lt;h2&gt;
  
  
  Why Does Open Innovation Matter?
&lt;/h2&gt;

&lt;p&gt;For me, this is the most important part of Bug Exorcist.&lt;/p&gt;

&lt;p&gt;Developers frequently work with private repositories, personal projects, credentials, and proprietary code. Sending that information to an external AI service isn't always desirable.&lt;/p&gt;

&lt;p&gt;By using an open-weight model with local inference, Bug Exorcist explores a different approach.&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;Privacy and control&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;The application communicates with a locally running model. Repository contents and debugging prompts don't need to be sent to a hosted AI API.&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;Offline capability&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;Once the model and application dependencies have been downloaded, the core AI workflow can operate without an internet connection.&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;Freedom to experiment&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;Developers can experiment with different compatible open-weight models without rebuilding the entire application around a proprietary AI provider.&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;Accessibility&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;Local inference avoids per-request charges from hosted AI APIs. Developers can experiment without depending on an external inference service, although they still need suitable hardware to run the model.&lt;/p&gt;

&lt;p&gt;I believe open innovation makes AI more accessible and gives developers greater control over how their tools work.&lt;/p&gt;

&lt;p&gt;For Bug Exorcist, that means debugging your own codebase with an AI model running on your own machine.&lt;/p&gt;

&lt;h2&gt;
  
  
  The Spirit of It
&lt;/h2&gt;

&lt;p&gt;Every developer has encountered a bug that seems impossible to explain.&lt;/p&gt;

&lt;p&gt;Sometimes it's a missing dependency. Sometimes it's an incorrectly initialized variable. And sometimes it's something hidden deep inside the codebase.&lt;/p&gt;

&lt;p&gt;I thought it would be fun to imagine these bugs as ghosts haunting our projects.&lt;/p&gt;

&lt;p&gt;Instead of another horror-themed chatbot, I wanted to give those ghosts a purpose: make them visible, investigate them, and help developers send them away.&lt;/p&gt;

&lt;p&gt;That's the spirit behind Bug Exorcist.&lt;/p&gt;

&lt;p&gt;Because sometimes, debugging feels less like programming and more like performing an exorcism. 😄&lt;/p&gt;

&lt;p&gt;Cast out bugs. Bring your code back to life. 👻&lt;/p&gt;

&lt;h2&gt;
  
  
  Prize Categories
&lt;/h2&gt;

&lt;p&gt;Gemma — Best Use of Gemma: Bug Exorcist uses Google's open-weight Gemma model as its local AI debugging engine through Ollama.&lt;/p&gt;

</description>
      <category>devchallenge</category>
      <category>hf26challenge</category>
    </item>
    <item>
      <title>OutStep: Turning Everyday Walks Into Adventures With Open-Weight AI</title>
      <dc:creator>Parth Ingle</dc:creator>
      <pubDate>Sun, 11 Oct 2026 18:27:59 +0000</pubDate>
      <link>https://dev.to/parth_ingle/outstep-turning-everyday-walks-into-adventures-with-open-weight-ai-32c4</link>
      <guid>https://dev.to/parth_ingle/outstep-turning-everyday-walks-into-adventures-with-open-weight-ai-32c4</guid>
      <description>&lt;p&gt;&lt;em&gt;This is a submission for the &lt;a href="https://dev.to/challenges/hacktoberfest-week1-2026-10-05"&gt;Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass&lt;/a&gt;&lt;/em&gt;&lt;/p&gt;

&lt;h2&gt;
  
  
  What I Built
&lt;/h2&gt;

&lt;p&gt;What if an AI application's success wasn't measured by how long you stayed on the screen, but by how quickly it helped you leave it?&lt;/p&gt;

&lt;p&gt;That's the idea behind &lt;strong&gt;OutStep&lt;/strong&gt;.&lt;/p&gt;

&lt;p&gt;I built &lt;em&gt;OutStep&lt;/em&gt; as a mobile-first outdoor companion that turns free time into personalized real-world missions. Instead of giving users another chatbot to talk to, it gives them something interesting to go and do.&lt;/p&gt;

&lt;p&gt;Users choose an activity, their available time, energy level, and difficulty. OutStep then generates an outdoor mission with a title, a clear objective, individual tasks, and estimated durations.&lt;/p&gt;

&lt;p&gt;The goal is to make going outside feel like an adventure rather than another item on a productivity checklist.&lt;/p&gt;

&lt;p&gt;A mission might turn an ordinary outdoor space into a Nature Bingo challenge or ask the user to become a visual detective, searching for overlooked details in the world around them.&lt;br&gt;
But generating the mission is only the beginning.&lt;/p&gt;

&lt;p&gt;OutStep also has an activity-tracking flow where users can start tasks, mark them complete, skip activities, continue past their target time, and review their progress afterward.&lt;/p&gt;

&lt;p&gt;Running late doesn't automatically mean failure. Time is a guide, not a punishment.&lt;/p&gt;

&lt;p&gt;My favourite part of this approach is that the AI doesn't need to remain active throughout the activity. It creates the plan, and then the application takes over.&lt;/p&gt;

&lt;p&gt;The AI plans the adventure. The app tracks the adventure. The human experiences the adventure.&lt;/p&gt;

&lt;h2&gt;
  
  
  Demo
&lt;/h2&gt;

&lt;p&gt;Try OutStep: &lt;a href="https://outstep.onrender.com/" rel="noopener noreferrer"&gt;OutStep&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;The deployed prototype lets you generate an outdoor mission and try the activity-tracking experience.&lt;/p&gt;

&lt;h2&gt;
  
  
  Code
&lt;/h2&gt;

&lt;p&gt;GitHub repository: &lt;a href="https://github.com/parthingle-codes/outstep" rel="noopener noreferrer"&gt;Go To GitHub&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;The repository contains the frontend, backend, mission-generation logic, tests, and deployment configuration.&lt;/p&gt;

&lt;h2&gt;
  
  
  How I Built It
&lt;/h2&gt;

&lt;p&gt;I built OutStep using React, Vite, Python, and FastAPI, with OpenAI's GPT-OSS 20B open-weight model powering mission generation through NVIDIA's hosted inference API.&lt;/p&gt;

&lt;p&gt;I used OpenCode, an AI coding agent, to help develop features, improve mission creativity, and debug the application.&lt;/p&gt;

&lt;p&gt;The backend uses Pydantic to validate structured missions, while deterministic application logic handles task timers, completion, skipped tasks, and summaries without repeated AI calls.&lt;/p&gt;

&lt;p&gt;One challenge was overcoming repetitive, generic missions. I introduced mission archetypes such as Nature Bingo, Visual Detective, and Pattern Breaker to encourage more varied outdoor adventures.&lt;/p&gt;

&lt;p&gt;GPT-OSS's Apache 2.0 license gives the project flexibility to explore self-hosting and alternative inference setups in the future. Currently, AI generation uses NVIDIA's hosted service, while saved missions can continue without repeated AI requests.&lt;/p&gt;

&lt;h2&gt;
  
  
  Why Does Open Innovation Matter?
&lt;/h2&gt;

&lt;p&gt;For OutStep, open innovation matters because I wanted the core intelligence to be adaptable rather than inseparable from one proprietary model API.&lt;/p&gt;

&lt;p&gt;OutStep uses GPT-OSS 20B, an open-weight model released under the Apache 2.0 license, subject to its applicable usage policy.&lt;/p&gt;

&lt;p&gt;The availability of the weights creates options that a closed model API alone cannot provide. Developers can download the model, explore self-hosting, adapt it using compatible tools, and experiment with fine-tuning and alternative inference runtimes.&lt;/p&gt;

&lt;p&gt;For this first public version, I chose NVIDIA's hosted inference endpoint so that people can try OutStep without downloading a large model themselves.&lt;/p&gt;

&lt;p&gt;That choice also has a limitation: the deployed mission generator depends on an external inference service. I'm not claiming that the current public demo runs the model locally on every phone or can generate new missions without internet access.&lt;/p&gt;

&lt;p&gt;What I find valuable is the flexibility available for future development. The backend can be adapted to different compatible inference providers, while the activity engine remains independent of the model.&lt;/p&gt;

&lt;p&gt;That separation means the model can evolve without requiring the entire application to be redesigned.&lt;/p&gt;

&lt;p&gt;Open innovation, for me, is about more than publishing a repository. It's about being able to understand, adapt, and improve the technology at the core of a product.&lt;/p&gt;

&lt;p&gt;And OutStep explores an idea I find particularly interesting: AI doesn't always need to encourage more interaction. Sometimes its best contribution is to help us spend less time using it.&lt;/p&gt;

&lt;h2&gt;
  
  
  My Agent Session
&lt;/h2&gt;

&lt;p&gt;I used OpenCode as a coding agent during development to inspect the codebase, implement improvements, and run tests.&lt;br&gt;
I haven't captured a DevRelay session for this build, so there is no session link to embed.&lt;/p&gt;

&lt;h2&gt;
  
  
  Prize Categories
&lt;/h2&gt;

&lt;p&gt;Best Use of Render — OutStep is deployed on Render, which hosts the application's frontend and backend.&lt;/p&gt;

&lt;h2&gt;
  
  
  What's Next?
&lt;/h2&gt;

&lt;p&gt;OutStep is just the beginning. Here's what I'd like to explore next:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;strong&gt;Offline-first PWA:&lt;/strong&gt; Make missions easier to access outdoors, even with limited connectivity.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Discovery Journal:&lt;/strong&gt; Let users record observations and personal reflections, with optional photo evidence.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Smarter Missions:&lt;/strong&gt; Improve mission variety, personalization, and adaptability to different environments and activity levels.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Progress &amp;amp; Streaks:&lt;/strong&gt; Encourage consistent outdoor exploration without turning it into another screen-time habit.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Local AI Inference:&lt;/strong&gt; Explore running open-weight models locally to reduce reliance on hosted inference APIs.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;The long-term goal is simple: &lt;strong&gt;use AI to get people away from their screens, not keep them glued to one.&lt;/strong&gt;&lt;/p&gt;

</description>
      <category>devchallenge</category>
      <category>hf26challenge</category>
    </item>
    <item>
      <title>Starting My Developer Journey</title>
      <dc:creator>Parth Ingle</dc:creator>
      <pubDate>Mon, 05 Oct 2026 11:54:31 +0000</pubDate>
      <link>https://dev.to/parth_ingle/starting-my-developer-journey-1lf5</link>
      <guid>https://dev.to/parth_ingle/starting-my-developer-journey-1lf5</guid>
      <description>&lt;p&gt;Hi!!&lt;br&gt;
I am Parth, a second-year Computer Science Engineering student, and I'm currently exploring different areas of software development and trying to learn by actually building things.&lt;/p&gt;

&lt;p&gt;This is the beginning of my DEV journey, and hopefully there will be many more posts, projects, bugs, and lessons along the way.&lt;/p&gt;

&lt;p&gt;Thanks for reading!&lt;/p&gt;

</description>
      <category>devto</category>
      <category>career</category>
    </item>
  </channel>
</rss>
