Full-stack Tanstack & FastAPI

An interactive 3D space where thousands of movies and novels are mapped as vector embeddings. The UI lets you freely explore, search, and compare titles with smooth tools for navigation and discovery.

Chrome extension

A distraction-blocking Chrome extension that lets you block sites, use Pomodoro timers, get focus reminders, track your focus time, and use a stripped-down YouTube mode that hides recommendations. It also includes quick access to lofi music.

AI Assistant Generator

A Retrieval-Augmented Generation chatbot that uses user-provided files and links as context. It handles PDFs, Office documents, spreadsheets, text, URLs, YouTube videos, and audio. It also exposes API endpoints where users can register resources so that any downstream chat requests can access them.

Neural Network Implementation

A minimal Python project implementing an autograd engine and multilayer perceptron (MLP) from scratch. It builds computation graphs using a Value class, supports backpropagation, and trains a small neural network with manual gradient descent on a toy dataset.

Capstone Project

A real-time ASL-to-text system that turns video of sign language into written captions. We built and trained our own LSTM model for gesture recognition, then integrated it into a custom video-calling platform so ASL users can have live captions during calls.

Software Engineer Intern

Migrated core customer service application from SOAP to RESTful APIs. Refactored the Java/Swing desktop client to consume JSON payload from REST endpoints, rewriting service classes and data models. Replaced SOAP calls with RESTful endpoints in the Spring Boot server.

Machine Learning Model

Constructed a Convolutional Neural Network (CNN) from scratch and trained it for sentiment analysis on textual content, utilizing a dataset sourced from the Reddit website.

Research/Application

Developed the Rush Hour puzzle game in Python, incorporating pathfinding algorithms such as A*, Greedy Best-First Search (GBFS), and Uniform Cost Search (UCS). The objective was to determine the optimal path to the goal state from a board configuration provided as a text file.

Database application + Web Application

Designed and implemented a Hospital Employee Status System, ensuring accurate tracking of employee health using MySQL and custom SQL triggers. This also included a front end interface using PHP, and a python flask scheduler to send emails.

Web Application

Developed a web application that facilitates the exchange of services between users. The application was built with Next.js and deployed on Google Cloud.

CPU Design for FPGA

Designed a CPU in VHDL, based on the ARM architecture and incorporated all core components such as the ALU, registers, and sign extension. The design was first simulated for correctness using ModelSim, then synthesized and implemented on an FPGA board.

Android application

Developed an Android application that enables users to control HVAC system remotely, also allowing users to monitor temperatures at specific locations and offering recommendations for temperature management.