I’ve built a full-stack AI-powered chat application that interacts with PDF documents, offering real-time insights and efficient data retrieval. This solution combines secure authentication, modern front-end frameworks, and back-end integrations with Google Firebase that stores vector embeddings to interact with Microsoft Azure LLM API.
Tech Stack Highlights:
• Next.js for server-side rendering and optimal performance
• Clerk for secure, streamlined authentication
• Tailwind CSS for rapid UI development and responsive design
• TypeScript for strong typing and maintainable code
• Google Firebase to store vector embeddings generated from the PDF
• Vercel to deploy and scale the web app
• GPT4-O API using Microsoft Azure functions
AI-Chat with PDF
1.
2.
Easy-Cook
I collaborated with a team of five classmates to build a feature-rich cooking and recipe website, with dual modes of interfaces dedicated to the elderly and people with disabilities. It is a project that is user-centric design, and streamlined in functionality. Our efforts combined modern web standards and best practices to deliver an engaging platform, and I was honored to be rated top of the CS3483 class at the end of the semester.
Tech Stack Highlights:
• JavaScript for dynamic content and interactive features
• Plain HTML5 for structured page layouts
• CSS for responsive, visually appealing design
• Rapid prototyping with Figma UI/UX design samples
• Github to deploy the website
Machine Learning & Data Science
My machine learning and data science endeavor involves intense labor that I had to give for developing my own deep learning models using TensorFlow, as well as the use of Scikit Learn to handle crucial tasks such as data preprocessing, feature engineering, and model evaluation, ensuring robust and scalable solutions.
1.
Bitcoin Prediction Model
I’ve developed a TensorFlow-based time-series machine learning model that forecasts Bitcoin’s performance in the stock market, all within Jupyter Notebook. By leveraging deep learning techniques and data-driven insights, this project sheds light on market patterns for better investment decision-making.
Tech Stack Highlights:
• TensorFlow for building robust deep learning architectures
• Jupyter Notebook for an interactive data exploration workflow
• Time-series analysis focusing on Bitcoin’s historical data
• Data preprocessing, feature engineering, and hyperparameter tuning
• Interpretability and performance evaluation for more accurate predictions
2.
Heart-Disease Prediction Model
I’ve developed a Sci-Kit Learn–based Heart Disease Prediction model that leverages powerful machine learning techniques to identify early warning indicators. By cleaning and preprocessing medical data, training robust classifiers, and analyzing performance metrics in Jupyter Notebook, this project provides a practical path towards enhanced clinical decision-making.
Tech Stack Highlights:
• Sci-Kit Learn for model development and evaluation
• Jupyter Notebook for interactive workflows and data visualization
• Data preprocessing and feature engineering for optimal model accuracy
• Machine learning model pipeline implementation (training, validation, testing)
• Comprehensive performance analysis using metrics such as accuracy and precision
3.
Hand-Gesture Based ML Web App
I’ve built an AI-powered web application showcasing real-time hand and gesture tracking using the ML5 library, enabling seamless multimodal interactions for image manipulation. By integrating computer vision techniques, intuitive UI/UX design, and robust client-side scripting, this project demonstrates how emerging technologies can transform user experiences.
Tech Stack Highlights:
• ML5 library for AI-driven hand and gesture recognition
• JavaScript for dynamic content and real-time interactions
• HTML5/CSS for responsive layouts and appealing UI design