<?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: Rohit</title>
    <description>The latest articles on DEV Community by Rohit (@rohit_choudhary_9c5862ece).</description>
    <link>https://dev.to/rohit_choudhary_9c5862ece</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%2F3712356%2Fa122da29-7711-469c-973b-6c001d14484e.jpg</url>
      <title>DEV Community: Rohit</title>
      <link>https://dev.to/rohit_choudhary_9c5862ece</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/rohit_choudhary_9c5862ece"/>
    <language>en</language>
    <item>
      <title>COGNI-VANTA</title>
      <dc:creator>Rohit</dc:creator>
      <pubDate>Sat, 31 Oct 2026 18:30:02 +0000</pubDate>
      <link>https://dev.to/rohit_choudhary_9c5862ece/cogni-vanta-2b43</link>
      <guid>https://dev.to/rohit_choudhary_9c5862ece/cogni-vanta-2b43</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;I built &lt;strong&gt;Cognivanta&lt;/strong&gt;, an AI-powered learning platform that turns study notes into interactive learning experiences.&lt;/p&gt;

&lt;p&gt;The idea behind the project is simple: every set of notes contains ideas waiting to be connected. Cognivanta helps bring those ideas to the surface by transforming uploaded study material into structured concepts, explanations, concept relationships, and quiz questions. Instead of only rereading a page, learners can explore what it means, see how its ideas relate, and test what they remember.&lt;/p&gt;

&lt;p&gt;I interpreted &lt;strong&gt;“Ghost in the Machine”&lt;/strong&gt; as the hidden knowledge inside our everyday learning materials: the concepts and connections that are present in a page of notes but not immediately visible. Cognivanta uses AI to help reveal that “ghost” and turn it into something learners can explore.&lt;/p&gt;

&lt;p&gt;The project also includes interactive subject experiences for Biology, Physics, History, and Geography, alongside profile and learning-progress features.&lt;/p&gt;

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

&lt;p&gt;&lt;strong&gt;Video demo:&lt;/strong&gt; &lt;a href="https://youtu.be/22bTe2ij43c?si=3JTUehrf7DtWUg-h" rel="noopener noreferrer"&gt;https://youtu.be/22bTe2ij43c?si=3JTUehrf7DtWUg-h&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;The video will walk through the core experience: uploading study material, generating learning content, exploring connected concepts, and trying an interactive subject world.&lt;/p&gt;

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

&lt;p&gt;&lt;strong&gt;GitHub repository:&lt;/strong&gt; Rohi56u/COGNI-VANTA &lt;a href="https://github.com/Rohi56u/COGNI-VANTA" rel="noopener noreferrer"&gt;https://github.com/Rohi56u/COGNI-VANTA&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;Cognivanta is built as a multi-service application, with a React and Vite frontend, an Express backend, and a Python FastAPI service for AI-assisted content generation.&lt;/p&gt;

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

&lt;p&gt;The frontend is built with React and Vite. It provides the learning workspace, upload flow, concept visualizations, quizzes, subject experiences, and profile views.&lt;/p&gt;

&lt;p&gt;The backend uses Node.js, Express, and SQLite for authentication, document handling, learning data, and progress-related features. A separate Python FastAPI service processes uploaded material and connects it to Gemini for AI-assisted transcription and concept generation.&lt;/p&gt;

&lt;p&gt;The AI workflow is designed to turn study material into structured learning content, including concepts, relationships, and recall questions. The repository also includes a sample-data path for development and demonstration.&lt;/p&gt;

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

&lt;p&gt;Cognivanta is open source because learning tools should be open to inspection, adaptation, and improvement. Developers and educators can explore how the application is put together, adapt the learning experience for different subjects, and contribute ideas or features of their own.&lt;/p&gt;

&lt;p&gt;The project currently uses Gemini’s API for AI generation, so I’m not presenting the model itself as open source. Open innovation still matters here: the surrounding application, learning flows, and integration are available for others to study and build on. Keeping those parts open also makes it easier to experiment with alternative models and providers over time.&lt;/p&gt;

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

&lt;p&gt;[Optional: Add your DevRelay session link or embed here.]&lt;/p&gt;

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

&lt;p&gt;A page of notes can look like a collection of disconnected facts. But behind it are ideas, patterns, and questions a learner may not have noticed yet.&lt;/p&gt;

&lt;p&gt;Cognivanta is my attempt to make that hidden layer visible: to turn notes into connections, connections into questions, and questions into a more active way of learning.&lt;/p&gt;

</description>
      <category>devchallenge</category>
      <category>hf26challenge</category>
    </item>
    <item>
      <title>THE GHOST CONTRIBUTOR</title>
      <dc:creator>Rohit</dc:creator>
      <pubDate>Sat, 24 Oct 2026 18:00:59 +0000</pubDate>
      <link>https://dev.to/rohit_choudhary_9c5862ece/the-ghost-contributor-c81</link>
      <guid>https://dev.to/rohit_choudhary_9c5862ece/the-ghost-contributor-c81</guid>
      <description>&lt;p&gt;&lt;em&gt;This is a submission for the &lt;a href="https://dev.to/challenges/hacktoberfest-week3-2026-10-19"&gt;Hacktoberfest Open-Source AI Challenge Week 3: Slow and Steady&lt;/a&gt;&lt;/em&gt;&lt;/p&gt;

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

&lt;p&gt;A missing docstring is easy to leave for later. The next person reading that function still has to work out what it does.&lt;/p&gt;

&lt;p&gt;I built &lt;em&gt;The Ghost Contributor&lt;/em&gt;, a local AI assistant that turns small documentation gaps into changes a developer can review and approve.&lt;/p&gt;

&lt;p&gt;You register a Git repository and choose an installed Ollama model. Ghost finds a supported documentation gap, drafts a focused suggestion, and presents the exact diff alongside quotations from the source. Approval creates a commit on a separate &lt;code&gt;ghost/…&lt;/code&gt; branch while your current branch stays in place.&lt;/p&gt;

&lt;p&gt;The current MVP handles missing docstrings in supported Python functions and explicitly marked, empty Markdown sections. It also supports scheduled runs and pause/resume controls. Remote publishing is a separate, explicit action.&lt;/p&gt;

&lt;p&gt;It is designed for developers and open-source maintainers who want help with recurring documentation chores while keeping the final decision.&lt;/p&gt;

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

&lt;p&gt;&lt;em&gt;45-second video:&lt;/em&gt;&lt;br&gt;
&lt;a href="https://youtu.be/HI0cluBc5Eg?si=of-yp4syhWQU_eNj" rel="noopener noreferrer"&gt;&lt;/a&gt; &lt;br&gt;
&lt;a href="https://youtu.be/HI0cluBc5Eg?si=of-yp4syhWQU_eNj" rel="noopener noreferrer"&gt;https://youtu.be/HI0cluBc5Eg?si=of-yp4syhWQU_eNj&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;The video uses screenshots from a real local &lt;code&gt;qwen2.5-coder:7b&lt;/code&gt; run: source analysis, a generated proposal, evidence review, approval, and a successful Git commit. Motion graphics explain the workflow; the inference wait is compressed and labelled in the video.&lt;/p&gt;

&lt;p&gt;Ghost runs on the developer’s computer, with a browser dashboard connected to the local application and Ollama.&lt;/p&gt;

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

&lt;p&gt;&lt;em&gt;Source repository:&lt;/em&gt;&lt;br&gt;
&lt;a href="https://github.com/Rohi56u/Ghost-contributer" rel="noopener noreferrer"&gt;&lt;/a&gt;&lt;br&gt;
&lt;a href="https://github.com/Rohi56u/Ghost-contributer" rel="noopener noreferrer"&gt;https://github.com/Rohi56u/Ghost-contributer&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;The application is released under the &lt;strong&gt;MIT license&lt;/strong&gt;. The source includes the Python backend, dashboard, tests, Windows setup launchers, architecture notes, and verification records.&lt;/p&gt;

&lt;p&gt;Start with &lt;code&gt;README.md&lt;/code&gt;, or &lt;code&gt;START-HERE-WINDOWS.md&lt;/code&gt; for the Windows walkthrough. The recorded run is documented in &lt;code&gt;docs/REAL-RUN.md&lt;/code&gt;.&lt;/p&gt;

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

&lt;p&gt;The stack is Python, Ollama, the open-weight Qwen2.5-Coder model, LibCST, SQLite, Git worktrees, and a local HTML/CSS/JavaScript dashboard. I developed the project with AI coding assistance.&lt;/p&gt;

&lt;p&gt;Each component has a specific job:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;
&lt;em&gt;Ollama and Qwen2.5-Coder&lt;/em&gt; generate a documentation suggestion from selected source context.
-&lt;em&gt;Python validation&lt;/em&gt; checks the structured response, allowed target, and exact source quotations.
-&lt;em&gt;LibCST and syntax checks&lt;/em&gt; apply the supported Python insertion and check that executable code outside the docstring remains unchanged.
-&lt;em&gt;SQLite&lt;/em&gt; records proposals, decisions, and run history.
-&lt;em&gt;Git worktrees&lt;/em&gt; isolate the proposed change and its eventual commit.
-&lt;em&gt;The dashboard&lt;/em&gt; puts the diff, rationale, and evidence together for human review.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;One useful failure shaped version 0.2.1. Early model responses copied display line numbers into their evidence quotes, so the validator rejected them. I changed the prompt to send line numbers and source text as separate JSON fields. The exact-quote checks stayed strict, and the subsequent real-model run produced valid citations.&lt;/p&gt;

&lt;p&gt;Verification includes &lt;strong&gt;93 passing automated tests&lt;/strong&gt; and a real Linux run through dashboard approval and Git commit. Windows setup instructions are included; execution on Windows and macOS still needs verification. Exact quotations establish where a suggestion came from; a person still needs to judge whether the prose is correct.&lt;/p&gt;

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

&lt;p&gt;Documentation assistance needs source-code context. Running an open-weight model through local Ollama lets the inference happen on a machine the developer controls, with cloud features disabled.&lt;/p&gt;

&lt;p&gt;After dependencies and model weights are installed, the inference path uses the local server. There is no hosted-model API key or per-request API bill, although local hardware and electricity still have a cost.&lt;/p&gt;

&lt;p&gt;Developers can select another compatible installed model, inspect the prompts and validators, and change the automation’s behavior. That makes the project adaptable to different machines and repositories.&lt;/p&gt;

&lt;p&gt;Compared with sending source context to a hosted model API, this gives the maintainer direct control over where generation happens. Releasing the application under MIT also lets others inspect, improve, and reuse the maintenance workflow.&lt;/p&gt;

&lt;h2&gt;
  
  
  The Lazy Payoff
&lt;/h2&gt;

&lt;p&gt;In this week’s sample run, Ghost found the missing documentation for &lt;code&gt;retry_delay&lt;/code&gt;, generated a paragraph with four exact source quotations, and prepared a reviewable diff. After approval, it created the separate branch commit and left &lt;code&gt;main&lt;/code&gt; unchanged.&lt;/p&gt;

&lt;p&gt;The dashboard run action took &lt;strong&gt;47.633 seconds&lt;/strong&gt;, including generation and validation. That is an observed runtime for one example—not a measurement of time saved, and it excludes the later review and approval.&lt;/p&gt;

&lt;p&gt;The intended payoff is fewer repetitive steps: finding the gap, drafting the first paragraph, and preparing the proposed change. The maintainer can focus on the review. I have not yet measured savings against a manual baseline or tested sustained use across multiple repositories.&lt;/p&gt;

</description>
      <category>devchallenge</category>
      <category>ai</category>
      <category>opensource</category>
      <category>hf26challenge</category>
    </item>
  </channel>
</rss>
