Developed and launched an AI-powered retail shopping platform featuring barcode scanning, personalized recommendations, checkout, real-time inventory synchronization, and store analytics. Built authentication, payment processing, role-based access control, and recommendation systems using matrix factorization.

Built a lineage-based columnar database management system from scratch in Python, featuring multi-version concurrency control, transaction rollback, primary-key indexing, and base/tail page architecture. Implemented an LRU buffer pool, disk persistence, and asynchronous merge operations without external database libraries.

Developed a computer vision interface that translates hand gestures into mouse controls with 95% accuracy. Supports web browsing, drawing, playing chess, and controlling games like Minecraft using real-time hand tracking.

Designed and trained a convolutional neural network to colorize grayscale images, achieving a 26% improvement over the baseline model. Built a Flask-based web interface for interacting with the model.

Developed a CNN-based dog breed classifier achieving 90% accuracy across 120 breeds using a dataset of over 20,000 images. Built an interactive Streamlit application for image classification.

Created a real-time facial emotion recognition application during HackDavis 2024. Developed a cross-platform Flutter interface connected to a Python FastAPI backend for processing facial expressions.

Built a real-time American Sign Language recognition application during HackDavis 2025, translating over 30 ASL letters and common gestures using computer vision and machine learning.

Developed an AI-powered focus companion during HackDavis 2026 that tracks browser activity, identifies distraction patterns, and displays productivity insights. Built as a browser extension with a React frontend, FastAPI backend, Gemini integration, and MongoDB.

Engineered a portable IoT pedometer for real-time step tracking and GPS route mapping using accelerometer data. Integrated a CC3200 microcontroller with AWS IoT Core and Lambda for remote monitoring, an OLED display, and Google Maps visualization. Implemented I2C, UART, GPIO, SPI, and HTTP communication.

Led a four-person team to develop and publish a two-player platformer in Godot. Showcased the game to over 200 students and faculty, receiving positive feedback from judges and an invitation to an end-of-year exhibition.

Created a top-down tower defense clicker game featuring gravity-based mechanics and an orbital system during a game jam. Developed the gameplay and interactive systems using Godot and GDScript.