Machine Learning & Full-Stack Developer
I'm a passionate developer with experience in machine learning, natural language processing, and full-stack web development. I enjoy building scalable applications and tackling challenging problems with innovative solutions. This portfolio showcases some of my key projects. Currently persuing my B.Tech. from IIT Madras.
Programming Languages
Frontend Development
Backend Development
Machine Learning - Data Science
Tools
-
Description: A modern, full-stack real-time chat application built with the MERN stack, emphasizing a responsive user interface and core communication features.
-
Highlights:
- Real-time Communication: Utilizes Socket.io for instant messaging capabilities.
- Responsive Design: Provides a seamless user experience across different devices using Tailwind CSS and DaisyUI.
- Loading Skeletons: Implements loading skeletons to improve perceived performance during data fetching.
- Cloudinary Integration: Uses Cloudinary for efficient image storage and delivery.
- Modern Stack: Full MERN stack project
- Structured Development: Follows a phased development roadmap for organized feature delivery.
-
Technologies: MongoDB, Express.js, React.js, Node.js, Socket.io, Tailwind CSS, DaisyUI, Zustand, Cloudinary, JWT, bcryptjs
-
Description: An AI-powered framework for extracting key-value pairs and generating summaries from PDFs with exceptional accuracy. Designed to handle diverse document formats, from legal documents to research papers.
-
Highlights:
- High Accuracy: Achieves 99.22% accuracy in text extraction using Fitz.
- Advanced NLP: Leverages SpaCy for Named Entity Recognition (NER) and regular expressions for specialized parsing (e.g., legal dates).
- Reinforcement Learning Optimization: Employs reinforcement learning to improve parsing adaptability for dynamic PDFs and enables retraining with custom datasets.
- Structured Data Output: Transforms unstructured PDF content into structured, usable data.
-
Technologies: SpaCy, Fitz, RegEx, Reinforcement Learning, Python
-
Description: An LLM-powered tool that allows users to extract information, transcribe, and ask questions about YouTube video content. Provides a seamless way to interact with video transcripts.
-
Highlights:
- Speech-to-Text: Utilizes OpenAI Whisper for high-quality speech-to-text conversion.
- Audio Extraction: Uses FFMPEG for efficient audio extraction from YouTube videos.
- NLP-Driven Querying: Employs Hugging Face Transformers for natural language processing and query resolution.
- Scalable Design: Built with a scalable architecture to support multilingual transcription, advanced summarization, and AI-powered query handling. Future-proofed for real-time processing and cloud deployment.
-
Technologies: OpenAI Whisper, Hugging Face Transformers, FFMPEG, Python
- Description: A full-stack MERN application simulating a food delivery service, complete with secure authentication, separate customer and administrator panels, cart management, and Stripe payment integration.
- Highlights:
- Secure Authentication: Implements robust user authentication using JWT (JSON Web Tokens) and bcrypt for password hashing.
- RESTful API: Well-structured RESTful APIs built with Express.js and Node.js.
- Database Management: Uses MongoDB and Mongoose for efficient data modeling and CRUD operations.
- State Management: Leverages Redux Toolkit for predictable state management in the React frontend.
- Payment Integration: Integrates Stripe Webhooks for secure and reliable payment processing.
- Input Validation: Uses Validator for protecting data integrity.
- Technologies: MongoDB, Express.js, React, Node.js, Redux Toolkit, JWT, bcrypt, Stripe, Validator, Local Storage
-
Description: A RESTful API that enables users to track movies, TV shows, books, and anime, and manage personalized watchlists. Focuses on security and performance.
-
Highlights:
- Secure Authentication: Employs JWT, bcrypt, and helmet for robust authentication and security.
- Microservices Architecture: Designed with a modular microservices architecture for improved maintainability and scalability.
- Security Features: Includes rate limiting, input validation, and MongoDB sanitization to protect against common web vulnerabilities.
- API Functionality: Provides features for filtering, sorting, and pagination to enhance user experience.
-
Technologies: Node.js, Express.js, Mongoose, JWT, bcrypt, helmet, MongoDB
-
Description: A React-based note-taking web application inspired by Google Keep, allowing users to create, edit, and delete notes with a clean and intuitive user interface.
-
Highlights:
- Responsive UI: Built with React, JSX, and Material-UI for a dynamic and responsive user experience.
- State Management: Uses React state for efficient note management.
- Modern Development Practices: Supports CI/CD deployment with Render integration.
-
Technologies: React, JSX, Material-UI, React DOM, Render
- ML for Marine Autonomy (OCEANA IIT-Madras): Secured 3rd place. Developed a CNN-based Convolutional Autoencoder for efficient underwater image transmission and decoding. Achieved 85%+ reconstruction accuracy.
- Pravartak Datathon (Research Park, IIT-Madras): Secured 4th place. Built a hypertuned regression model to predict US house prices, achieving a 92% MSE reduction using techniques like EDA, spatial analysis (GeoPandas, Matplotlib), and feature engineering.
I'm always open to collaboration and new opportunities. Feel free to reach out!