<?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: Divyansh Singh</title>
    <description>The latest articles on DEV Community by Divyansh Singh (@divyyansh_063).</description>
    <link>https://dev.to/divyyansh_063</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%2F4158551%2F6bae3a41-f87c-4f77-bc05-f9048aff9a34.png</url>
      <title>DEV Community: Divyansh Singh</title>
      <link>https://dev.to/divyyansh_063</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/divyyansh_063"/>
    <language>en</language>
    <item>
      <title>LocalLedger AI</title>
      <dc:creator>Divyansh Singh</dc:creator>
      <pubDate>Sat, 17 Oct 2026 06:47:43 +0000</pubDate>
      <link>https://dev.to/divyyansh_063/localledger-ai-2g2h</link>
      <guid>https://dev.to/divyyansh_063/localledger-ai-2g2h</guid>
      <description>&lt;h1&gt;
  
  
  LocalLedger AI: Turn Receipts into Expenses with Local AI
&lt;/h1&gt;

&lt;p&gt;&lt;em&gt;This is a submission for the &lt;a href="https://dev.to/challenges/hacktoberfest-week2-2026-10-12"&gt;Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle&lt;/a&gt;&lt;/em&gt;&lt;/p&gt;

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

&lt;p&gt;Meet &lt;strong&gt;LocalLedger AI&lt;/strong&gt; — a privacy-focused receipt and invoice expense tracker powered by locally running AI.&lt;/p&gt;

&lt;p&gt;Keeping track of expenses shouldn't mean manually entering every transaction. Students, freelancers, and small businesses often collect receipts but struggle to turn them into organized financial records.&lt;/p&gt;

&lt;p&gt;LocalLedger AI simplifies that workflow. Users upload a receipt, let AI extract important transaction details, review and correct the results, and save the verified expense to a local database.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Features include:&lt;/strong&gt;&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;AI-powered receipt and invoice data extraction.&lt;/li&gt;
&lt;li&gt;An editable review step before saving expenses.&lt;/li&gt;
&lt;li&gt;Local SQLite storage for expense records.&lt;/li&gt;
&lt;li&gt;Spending analytics by category and month.&lt;/li&gt;
&lt;li&gt;CSV export for further analysis.&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;The long-term goal is to make this a lightweight expense-management tool for people who want to spend less time entering data and more time understanding where their money goes.&lt;/p&gt;

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

&lt;p&gt;&lt;strong&gt;Demo video:&lt;/strong&gt; &lt;a href="https://youtu.be/PSt0bdE1pNE" rel="noopener noreferrer"&gt;https://youtu.be/PSt0bdE1pNE&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;The demo will show the complete workflow: uploading a receipt, extracting its details, reviewing the results, saving the transaction, viewing spending analytics, and exporting the data.&lt;/p&gt;

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

&lt;p&gt;&lt;strong&gt;GitHub repository:&lt;/strong&gt; &lt;a href="https://github.com/divyyansh-063/LocalLedgerAI" rel="noopener noreferrer"&gt;https://github.com/divyyansh-063/LocalLedgerAI&lt;/a&gt;&lt;/p&gt;

&lt;p&gt;The project is built with Python, Flask, SQLite, HTML, CSS, JavaScript, and Ollama.&lt;/p&gt;

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

&lt;p&gt;The application uses &lt;strong&gt;Gemma 3 (&lt;code&gt;gemma3:4b&lt;/code&gt;) through Ollama&lt;/strong&gt; to power the receipt-extraction workflow locally.&lt;/p&gt;

&lt;p&gt;I built the application around a simple pipeline:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;
&lt;strong&gt;Upload:&lt;/strong&gt; The user provides a receipt or invoice image.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Extract:&lt;/strong&gt; Gemma processes the receipt and extracts transaction details.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Review:&lt;/strong&gt; The user checks and corrects the extracted information.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Save:&lt;/strong&gt; Verified data is stored in SQLite.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Analyze:&lt;/strong&gt; The dashboard summarizes expenses by category and month.&lt;/li&gt;
&lt;li&gt;
&lt;strong&gt;Export:&lt;/strong&gt; Users can download transaction records as a CSV file.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;Flask connects the backend workflow to the web interface, while SQLite stores expense records.&lt;/p&gt;

&lt;p&gt;An important design decision is keeping the user involved in the process. AI-extracted information can be imperfect, so users can review the result before it becomes a saved expense.&lt;/p&gt;

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

&lt;p&gt;Financial documents can contain sensitive information, and not everyone wants to send them to an external AI service.&lt;/p&gt;

&lt;p&gt;Open-weight models such as Gemma make it possible to experiment with AI-powered workflows locally, giving developers more control over inference and data handling.&lt;/p&gt;

&lt;p&gt;Building LocalLedger AI with Gemma and Ollama also gave me an opportunity to explore how open AI models can be integrated into a practical application instead of being used only for a chatbot.&lt;/p&gt;

&lt;p&gt;Open innovation makes these tools more accessible to students and independent developers who want to build, inspect, adapt, and extend AI-powered products.&lt;/p&gt;

&lt;p&gt;Local inference can also reduce reliance on per-request hosted AI fees, although hardware and electricity costs still matter.&lt;/p&gt;

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

&lt;ul&gt;
&lt;li&gt;Best Use of Gemma&lt;/li&gt;
&lt;/ul&gt;

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

&lt;p&gt;I want to improve duplicate receipt detection, transaction filtering, extraction validation, and budgeting features to make LocalLedger AI more reliable for everyday use.&lt;/p&gt;

&lt;p&gt;Building this project has been an opportunity to explore local AI, open-source development, and the process of turning a practical problem into a working application.&lt;/p&gt;

&lt;h1&gt;
  
  
  hf26challenge #Gemma
&lt;/h1&gt;

</description>
      <category>devchallenge</category>
      <category>hf26challenge</category>
    </item>
    <item>
      <title>Microclimate Runner</title>
      <dc:creator>Divyansh Singh</dc:creator>
      <pubDate>Sun, 11 Oct 2026 13:35:06 +0000</pubDate>
      <link>https://dev.to/divyyansh_063/microclimate-runner-578d</link>
      <guid>https://dev.to/divyyansh_063/microclimate-runner-578d</guid>
      <description>&lt;h2&gt;
  
  
  Microclimate Runner: A Local-AI Route Planner That Gets You Outside
&lt;/h2&gt;

&lt;p&gt;This is a submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass&lt;/p&gt;

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

&lt;p&gt;Microclimate Runner is an open-source running route planner designed to help people stop staring at route maps and actually go outside.&lt;/p&gt;

&lt;p&gt;The app uses your current location, target distance, and preferred priority to find nearby routes. It looks for routes with more shade, lower pollen exposure, and better overall comfort.&lt;/p&gt;

&lt;p&gt;The flow is intentionally short:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;Choose a distance and route preference.&lt;/li&gt;
&lt;li&gt;Let the app plan a nearby route.&lt;/li&gt;
&lt;li&gt;Read or listen to the briefing.&lt;/li&gt;
&lt;li&gt;Start Run Mode and leave the screen behind.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;While running, the app tracks progress with GPS and displays only the essential information needed to keep moving.&lt;/p&gt;

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

&lt;ul&gt;
&lt;li&gt;Live demo: &lt;a href="https://microclimate-runner.vercel.app/" rel="noopener noreferrer"&gt;https://microclimate-runner.vercel.app/&lt;/a&gt;
&lt;/li&gt;
&lt;/ul&gt;

&lt;h3&gt;
  
  
  Demo Notes
&lt;/h3&gt;

&lt;p&gt;The local development setup uses Gemma through Ollama for route ranking. The hosted Vercel demo uses the application's explicit deterministic fallback when local Ollama inference is unavailable, because a Vercel deployment cannot access Ollama running on my personal computer.&lt;/p&gt;

&lt;p&gt;The interface reports the scoring source instead of claiming that the hosted version used Gemma when it did not.&lt;/p&gt;

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

&lt;ul&gt;
&lt;li&gt;GitHub repository: &lt;a href="https://github.com/divyyansh-063/microclimate-runner" rel="noopener noreferrer"&gt;https://github.com/divyyansh-063/microclimate-runner&lt;/a&gt;
&lt;/li&gt;
&lt;/ul&gt;

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

&lt;p&gt;The project is built with Next.js, React, TypeScript, OpenStreetMap data, Open-Meteo weather and air-quality data, and a Mastra workflow.&lt;/p&gt;

&lt;p&gt;The open-source AI component is local Gemma running through Ollama. Gemma receives route candidates, runner preferences, and environmental conditions, then ranks the candidate paths for comfort.&lt;/p&gt;

&lt;p&gt;The main workflow is:&lt;/p&gt;

&lt;ol&gt;
&lt;li&gt;OpenStreetMap/Overpass provides nearby footpaths and route candidates.&lt;/li&gt;
&lt;li&gt;Open-Meteo provides current temperature, UV, wind, and pollen data when available.&lt;/li&gt;
&lt;li&gt;Local Gemma ranks the routes using an inspectable prompt.&lt;/li&gt;
&lt;li&gt;A deterministic heuristic scorer is used as an explicit fallback if Ollama or Gemma is unavailable.&lt;/li&gt;
&lt;li&gt;Optional ElevenLabs narration creates a short audio briefing.&lt;/li&gt;
&lt;li&gt;Run Mode uses browser GPS so the runner can put the phone away.&lt;/li&gt;
&lt;/ol&gt;

&lt;p&gt;The application never pretends that local AI was used when it was unavailable. The interface reports whether the result came from:&lt;/p&gt;

&lt;ul&gt;
&lt;li&gt;&lt;code&gt;gemma-local&lt;/code&gt;&lt;/li&gt;
&lt;li&gt;&lt;code&gt;heuristic-fallback&lt;/code&gt;&lt;/li&gt;
&lt;/ul&gt;

&lt;p&gt;To run the project locally:&lt;/p&gt;



&lt;div class="highlight js-code-highlight"&gt;
&lt;pre class="highlight plaintext"&gt;&lt;code&gt;
bash
npm install
npm run dev

## Prize Categories

I am entering the overall Hacktoberfest Week 1: Touch Grass challenge.

The project uses local open-source AI through Gemma and Ollama. I am not claiming a partner category unless the challenge page lists a category that this project genuinely qualifies for.
&lt;/code&gt;&lt;/pre&gt;

&lt;/div&gt;

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