Portfolio
Embedded Files
TERRA - NASA
TERRA - NASA
AR Terrain Analysis & Route Planning
TERRA is an augmented reality terrain analysis system developed at NASA Glenn Research Center to support rover mobility and field research. Built for Meta Quest 3, it scans real-world terrain, reconstructs 3D surfaces, visualizes slope conditions directly in AR, and generates routes that balance terrain difficulty and travel distance using A* pathfinding. The system also supports real-time slope, incline, and distance measurements to make terrain assessment more intuitive and accessible.
Developed and tested with researchers in NASA Glenn’s SLOPE Lab.
Presented at the 2026 National GEM Consortium Annual Conference.
Chartie - NASA
Chartie - NASA
Conversational AI for Data Exploration
Chartie is a conversational AI system developed at NASA Glenn Research Center to help researchers explore rover experiment data in immersive environments. Built for VR/AR, it allows users to ask questions through voice or text, automatically identifies relevant datasets, performs statistical analysis, and generates charts and spoken responses. The system combines conversational AI, data visualization, and immersive computing to make scientific data exploration faster and more intuitive.
Developed in collaboration with researchers in NASA Glenn’s SLOPE Lab.
Presented at the 2026 National GEM Consortium Annual Conference.
Voice Bloom
Voice Bloom
Adaptive Communication System
Voice Bloom is an adaptive, multilingual communication platform developed in collabaration with UC Riverside and UC Berkeley to support non-speaking and low-literacy users. Built with the ACE Lab, it combines context-aware phrase prediction, animated avatars, multilingual text-to-speech, and personalized visual prompts to make communication more intuitive and expressive. The system uses user context and communication patterns to surface relevant suggestions, reduce cognitive load, and support greater independence.
Developed with INSPIRE Lab (UCR) and ACE Lab (EECS, Berkeley).
Presented at the CITRIS Tech Policy Symposium: Data, Democracy, and Design.
VisionCart
VisionCart
VisionCart: Multi-agent E-Commerce
VisionCart is a multi-agent AI system developed at UC Berkeley to turn visual inspiration into personalized product recommendations. It analyzes vision boards and images to identify style preferences, generate search strategies, retrieve relevant products, and rank results based on aesthetic alignment. The project combines multimodal AI, agentic workflows, and product discovery to move beyond traditional keyword-based shopping.
Developed for UC Berkeley’s INFO 290 Generative AI course.
Awarded Best Poster at the INFO 290 Generative AI Poster Presentations.
Eyegentic Research Portal
Eyegentic Research Portal
AI Research Discovery Platform
Eyegentic is a RAG-based research platform developed with the Foundation Fighting Blindness to help researchers explore historical grant and scientific data. The system combines document ingestion, embeddings, vector search, and LLM-based synthesis to retrieve relevant information and generate grounded answers from large collections of research materials. The project also included stakeholder interviews, product requirements, UX design, and system architecture to make research discovery more efficient and accessible.
Developed in collaboration with the Foundation Fighting Blindness.
Used internally to evaluate research outcomes and support funding decisions.
Amodal Object Detection
Amodal Object Detection
Computer Vision for Occluded Object Detection
This project explored how computer vision systems can recognize objects even when they are partially hidden. Developed during LLNL’s 2025 Data Science Challenge, our cross-UC team trained U-Net and ConvLSTM models in PyTorch for amodal segmentation, RGB mask prediction, and video-based object tracking. The system was tested on synthetic and real-world driving data to improve detection of occluded pedestrians and vehicles.
Developed with a cross-UC research team at Lawrence Livermore National Laboratory.
Achieved an F1 score of 0.84 and IoU of 0.79.
Presented to LLNL leadership and faculty as part of the 2025 Data Science Challenge.
Project Navigator
Project Navigator
AI-Powered Project Management Assistant
Project Navigator is an AI-powered project management tool developed at UC Riverside to help teams generate plans, documentation, and project guidance through natural conversation. Built with the OpenAI API, Node.js, HTML, and CSS, the platform creates context-aware project materials based on user needs and ongoing conversations. I led a five-person team through the project’s research, development, and presentation.
Supported by a UCR undergraduate research mini-grant and more than $3,800 in scholarships.
Recipient of the Pillars of Excellence Award and semifinalist in the 2025 Mays Business School AI Pitch Competition.
Explore My GitHub
Explore My GitHub
Check out some of my featured projects that reflect my work in AI, data science, and educational technology:
Page updated
Google Sites
Report abuse