The latest articles on DEV Community by LAKSHITHAA V (@lakshiscoobyalt). https://dev.to/lakshiscoobyalt 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%2F4170804%2F965f991c-84cb-4044-a4be-ce4b8fef4c38.png https://dev.to/lakshiscoobyalt en LAKSHITHAA V Thu, 08 Oct 2026 09:27:00 +0000 https://dev.to/lakshiscoobyalt/campuslens-ai-a-gemma-4-multimodal-assistant-for-college-notices-3iip https://dev.to/lakshiscoobyalt/campuslens-ai-a-gemma-4-multimodal-assistant-for-college-notices-3iip <h1> CampusLens AI: A Gemma 4 Multimodal Assistant for College Notices </h1> <p>College students receive important information through notices, posters, assignment sheets, and other visual documents. Finding the important details manually can be time-consuming and confusing.</p> <p>For the MLH Hacktoberfest Hack Day Coimbatore, our team <strong>Synap Tech</strong> built <strong>CampusLens AI</strong>, a multimodal AI assistant that uses <strong>Gemma 4</strong> to understand college notices, extract important information, and answer questions about the document.</p> <h2> What We Built </h2> <p>CampusLens AI allows a student to upload a photo or image of a college notice.</p> <p>The system then:</p> <ol> <li>Understands the uploaded document using Gemma 4.</li> <li>Extracts important information such as: <ul> <li>Event</li> <li>Date</li> <li>Time</li> <li>Venue</li> <li>Deadline</li> <li>Requirements</li> </ul> </li> <li>Generates a concise summary.</li> <li>Allows the student to ask questions about the uploaded notice.</li> </ol> <p>For example, instead of manually reading a long event poster, a student can ask:</p> <blockquote> <p>"What do I need to bring?"</p> </blockquote> <p>or</p> <blockquote> <p>"When is the event?"</p> </blockquote> <p>and CampusLens AI provides an answer based on the uploaded document.</p> <h2> How We Built It </h2> <p>The main technologies used in our project are:</p> <ul> <li> <strong>Gemma 4</strong> for multimodal document understanding and question answering</li> <li> <strong>Google GenAI SDK</strong> for interacting with Gemma 4</li> <li> <strong>Python</strong> for the AI processing</li> <li> <strong>Streamlit</strong> for the web interface</li> <li> <strong>Git and GitHub</strong> for collaborative development</li> </ul> <p>Our core workflow is:</p> <p><strong>Upload Document → Gemma 4 → Understand → Extract Information → Ask Questions</strong></p> <p>We used prompt engineering to make the model return structured information and instructed it not to guess information that is not present in the document.</p> <h2> Why Gemma 4? </h2> <p>Gemma 4 was a great fit for this project because CampusLens needs to work with visual documents rather than only plain text.</p> <p>The multimodal capabilities allow the model to understand information directly from an uploaded notice and combine that understanding with natural-language question answering.</p> <h2> Challenges We Faced </h2> <p>During development, we encountered a few practical challenges.</p> <p>One challenge was model availability and stability. We initially tested <code>gemma-4-31b-it</code>, but experienced server errors on one of our development machines. We tested the available Gemma 4 models and switched our working application to <code>gemma-4-26b-a4b-it</code>, which worked reliably for our multimodal workflow.</p> <p>We also had to debug API connectivity, image processing, Streamlit integration, and Git collaboration issues while working as a team.</p> <p>Because the hackathon had a limited development window, we focused on making the core workflow reliable rather than adding too many extra features.</p> <h2> What We Learned </h2> <p>Through this project, we learned about:</p> <ul> <li>Multimodal AI and document understanding</li> <li>Gemma 4 integration</li> <li>Prompt engineering</li> <li>Building AI-powered interfaces with Streamlit</li> <li>API integration and debugging</li> <li>Collaborative Git and GitHub workflows</li> <li>Rapid MVP development during a hackathon</li> </ul> <h2> Future Scope </h2> <p>There are several features we would like to explore in the future:</p> <ul> <li>Calendar integration for extracted events</li> <li>Smart reminders for deadlines</li> <li>Action-item extraction</li> <li>Support for multiple document types</li> <li>Multilingual document understanding</li> <li>Searching across multiple uploaded notices</li> <li>Mobile-friendly improvements</li> </ul> <h2> Team </h2> <h3> Synap Tech </h3> <ul> <li> <strong>Lakshithaa V</strong> — AI/backend development, Gemma 4 integration, multimodal document processing, information extraction, Q&amp;A, Git/GitHub</li> <li> <strong>Pavitraa Surendran</strong> — UI and frontend development using Streamlit</li> <li> <strong>Shivamihit G</strong> — Documentation and presentation</li> <li> <strong>Rithvik Kumar</strong> — Prompt engineering, AI testing, and edge cases</li> </ul> <h2> Links </h2> <h3> GitHub Repository </h3> <p><a href="https://github.com/lakshiscooby-alt/campuslens-ai" rel="noopener noreferrer">https://github.com/lakshiscooby-alt/campuslens-ai</a></p> <h3> Demo Video </h3> <p><a href="https://youtu.be/EYpPQo6tEKM?si=82X_DOr5VlY8jvKz" rel="noopener noreferrer">https://youtu.be/EYpPQo6tEKM?si=82X_DOr5VlY8jvKz</a></p> <h2> Built For </h2> <p>This project was built during <strong>MLH Hacktoberfest Hack Day Coimbatore</strong>.</p> <p>We are particularly submitting CampusLens AI for the <strong>Best Use of Gemma 4</strong> challenge.</p> <p>Thanks to the organizers and mentors for giving us the opportunity to build and experiment with Gemma 4!</p> ai programming api machinelearning