The latest articles on DEV Community by Srushti Kaware (@povxsrushti). https://dev.to/povxsrushti 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%2F4176917%2F4d5c7b48-f5f9-4ea4-ade7-de390a0b7ea0.jpg https://dev.to/povxsrushti en Srushti Kaware Sat, 31 Oct 2026 17:47:51 +0000 https://dev.to/povxsrushti/built-haunted-code-reviewer-an-ai-powered-streamlit-app-entirely-from-my-phone-learned-cloud-1f03 https://dev.to/povxsrushti/built-haunted-code-reviewer-an-ai-powered-streamlit-app-entirely-from-my-phone-learned-cloud-1f03 ai buildinpublic hacktoberfest python Srushti Kaware Sat, 31 Oct 2026 17:43:53 +0000 https://dev.to/povxsrushti/haunted-code-reviewer-building-an-ai-code-reviewer-with-a-ghostly-twist-hacktoberfest-2026-3ih8 https://dev.to/povxsrushti/haunted-code-reviewer-building-an-ai-code-reviewer-with-a-ghostly-twist-hacktoberfest-2026-3ih8 <p>This is a submission for the "Hacktoberfest Open-Source AI Challenge Week 4: Ghost in the Machine" (<a href="https://dev.to/challenges/hacktoberfest-week4-2026-10-26">https://dev.to/challenges/hacktoberfest-week4-2026-10-26</a>)</p> <p><strong>What I Built</strong></p> <p>👻 Haunted Code Reviewer is an AI-powered code review web app that gives developers spooky but useful feedback.</p> <p>Users paste a code snippet, select a programming language, and summon a ghostly reviewer that identifies potential bugs, logic errors, unused variables, and code improvements.</p> <p>Inspired by the "Ghost in the Machine" theme, I gave the AI a hidden system prompt that makes it act like a ghost trapped inside a computer while still delivering accurate, beginner-friendly code reviews.</p> <p><strong>Demo</strong></p> <p>[Try Haunted Code Reviewer –<a href="https://haunted-code-reviewer.streamlit.app/" rel="noopener noreferrer">https://haunted-code-reviewer.streamlit.app/</a><br> ]</p> <p>Code</p> <p>[GitHub Repository–<a href="https://github.com/srushtikaware/Haunted-Code-Reviewer" rel="noopener noreferrer">https://github.com/srushtikaware/Haunted-Code-Reviewer</a> ]</p> <p><strong>How I Built It</strong></p> <p>I built the app using Python and Streamlit, with the Hugging Face Inference API connecting it to the open-weight Llama 3.1 8B Instruct model.</p> <p><strong>The app includes:</strong></p> <ul> <li>A code input area and programming language selector.</li> <li>A ghost-themed interface with custom CSS.</li> <li>An AI review system prompt that combines technical accuracy with spooky commentary.</li> <li>Secure API key management through Streamlit Secrets.</li> </ul> <p>I developed and deployed the project entirely from my phone using GitHub Codespaces, GitHub, and Streamlit Community Cloud.</p> <p><strong>Why Does Open Innovation Matter?</strong></p> <p>Open innovation made this project possible without expensive hardware or proprietary AI infrastructure.</p> <p>Using an open-weight model through Hugging Face allowed me to experiment with AI-powered code review using accessible tools. Open-source frameworks like Streamlit made it easy to build and share the application.</p> <p>This project showed me how open tools can lower barriers to development and help beginners participate in AI innovation.</p> <p><strong>The Spirit of It</strong></p> <p>Building this project entirely from my phone was a challenge of its own. From debugging indentation errors to troubleshooting cloud deployment, every obstacle taught me something new.</p> <p>The most exciting moment was seeing my own AI-powered app open in a browser after working through those challenges. It reminded me that creativity and persistence matter more than having expensive equipment.</p> <p>Prize Categories</p> <p>Open-Source AI Challenge</p> devchallenge hf26challenge