RAG with Gemma-3 is end-to-end and fully local system that lets users upload documents (PDF, TXT, MD, etc.) and chat with them using natural language. It runs on Gemma-3 LLM and mxbai-embeddings via Ollama, ensuring privacy and low-latency responses without needing the cloud. Built using LangChain, FastAPI, Streamlit, FAISS, Docker, and SQLite, it supports multi-file uploads, vector embeddings, real-time streaming responses, user authentication, and session history. Designed to be modular, clean, and fast - perfect for personal assistants, or educational use.

This is an AI-powered multilingual quiz generator that builds interactive quizzes from any news article or topic in any language. It detects the language, translates to English for LLM-based question generation via Gemini API, and translates the questions back to the original language. The platform supports MCQ, Multiple-Correct, True/False, and Numerical ans type questions, provides result evaluation, and features a performance analysis section. The app is deployed using Streamlit with a simple and intuitive interface.

A LangChain-powered QnA ChatBot, built with Streamlit. Supports real-time provider and model switching across OpenAI, Groq, Google, and Ollama. It dynamically lists models based on the selected LLM provider and maintains session history using advanced LangChain components.

LocalGPT is an offline clone of ChatGPT completely built from scratch. Uses open-source LLMs via Ollama and offers an interactive Streamlit-based interface. The tool supports conversation threads, real-time streaming responses, and multimodal inputs including vision models. It retains context per thread, allows model switching, and runs entirely without internet, making it ideal for local experimentation with LLMs.

The Distributed Attendance System extends a Smart Attendance System by enabling distributed video processing across multiple client machines. It supports both static and dynamic load balancing for real-time face recognition-based attendance marking. The web-based interface allows video uploads, and results are downloadable in Excel format. It ensures faster processing using distributed computing, all coordinated by a central Flask server.

This project is a real-time air quality and HVAC monitoring dashboard that collects data from various sensors for environmental factors via an Arduino and Raspberry Pi setup. It adheres to the Cloud-to-Fog-to-Things (C2F2T) model, ensuring low-latency, distributed processing. A local visualization is displayed on LCD-TFT screen. Also, The web app provides an interactive and responsive interface for real-time data visualization and emergency alerts, and uses Firebase for real-time data storage. Deployed on Vercel, it offers seamless access to both historical and live sensor data.

Ms Minutes is a standalone, autonomous voice assistant inspired by the Loki character. It responds to voice commands for weather, calculations, definitions, and general questions. The assistant also includes home automation capabilities, controlling smart appliances like a solenoid latch. Integrated with the OpenAI API and deployed on a Raspberry Pi Zero 2-W, the system functions independently and can be used anywhere with just a power adapter.

This project is a beginner-friendly travel guide application deployed fully on AWS infrastructure. It uses EC2 for hosting the Flask backend, RDS (MySQL) or SQLite for storing location data, S3 storage bucket for hosting public images and presents travel recommendations through a clean web interface. The project demonstrates basic cloud deployment, database integration, and responsive web design, making it ideal for understanding AWS fundamentals.

This project is incremental improvement of the previous travel guide application. Moving forward from the basic deployment, it now includes more AWS Services like RDS for database management, Cognito for user authentication, and SNS for notifications, Translate for content translation. The project is focused on utilizing multiple AWS services for scalability and security.

This simple Chrome extension enhances the VTOP (Chennai Campus) portal by dynamically adding a total marks row to each subject's marks table. Built with Manifest v3 and the Chrome Scripting API, it's lightweight, easy to install manually, and works seamlessly on Chromium-based browsers.

This project was developed to manage and verify entry for a food-based event organized by Swarajya: Marathi Literary Association at VIT Chennai. The system supports offline verification using MySQL, and a Streamlit-powered UI enables real-time entry marking and error correction. It features live dashboards showing marked/unmarked entries and a trend analysis chart to visualize footfall patterns. Designed for usability, it enables smooth operation across parallel queues using multiple devices on the same network.