SHIVAM RATHORE

Tech Stack: Python, OpenCV, MediaPipe, NumPy, Scikit-learn, Tkinter, NLP, ESP32-CAM

• Developed an AI-powered assistive communication system that enables patients with Locked-In Syndrome (LIS) to communicate using only eye movements.
• Designed the system to detect gaze direction, identify blinking patterns, and convert user intent into text and speech.
• Built an accessible user interface with large buttons and high-contrast elements for ease of use.

• Real-time eye and iris tracking using MediaPipe Face Mesh.
• Blink detection for click/selection operations.
• Virtual keyboard controlled entirely through eye movements.
• Predictive text suggestions to reduce typing effort.
• Text-to-Speech conversion for spoken communication.
• Custom phrase shortcuts for frequently used sentences.
• User calibration process to improve gaze accuracy.
• Notes section to save important messages.
• Responsive GUI suitable for differently-abled users.

• Used MediaPipe facial landmarks to accurately track iris positions.
• Processed webcam frames using OpenCV.
• Implemented gaze estimation based on normalized iris coordinates.
• Applied image preprocessing techniques to improve tracking stability.
• Optimized tracking to reduce false gaze detections.

• Integrated ESP32-CAM for wireless video streaming.
• Used Arduino Nano for hardware interfacing.
• Designed wearable smart-glasses prototype.
• Powered the system using a rechargeable battery and voltage regulator.

• Reduced noise caused by varying lighting conditions.
• Improved cursor stability during eye movement.
• Minimized accidental selections through blink validation.
• Increased responsiveness by optimizing frame processing.

• Computer Vision
• Human-Computer Interaction
• Python Development
• AI-assisted Accessibility
• Embedded Systems Integration
• Real-time Video Processing
• Problem Solving

Tech Stack: HTML, CSS, JavaScript, Tailwind CSS, Python, OpenAI API, Pinecone Vector Database

• Designed and developed an AI-powered note-taking web application inspired by Apple Notes and Google Keep.
• Integrated generative AI features to help users summarize, rewrite, expand, and improve notes.
• Built an intuitive and responsive interface focused on productivity.

• Create, edit, delete, and organize notes.
• AI-powered note summarization.
• Grammar correction and content enhancement.
• Text expansion and paraphrasing.
• Smart title generation.
• Responsive UI for desktop and mobile devices.
• Clean, distraction-free writing experience.

• Connected the application with the OpenAI API.
• Enabled intelligent rewriting and summarization of notes.
• Used prompt engineering to generate high-quality responses.
• Reduced manual editing effort through automated text improvements.

• Explored Pinecone Vector Database for semantic note storage.
• Generated embeddings for notes.
• Implemented semantic similarity search.
• Enabled retrieval of contextually similar notes rather than exact keyword matches.

• Built a responsive interface using Tailwind CSS.
• Implemented dynamic note rendering using JavaScript.
• Designed reusable UI components.
• Improved user experience with smooth interactions.

• Managed API integration and response handling.
• Optimized prompt design for consistent AI output.
• Structured note storage for future semantic retrieval.
• Handled asynchronous API requests efficiently.

• Full-Stack Web Development
• REST API Integration
• Prompt Engineering
• Vector Databases
• Frontend Development
• JavaScript
• UI/UX Design
• AI Application Development