I’m a Computer Science student at Texas State University, working at the intersection of software engineering, machine learning, and computer vision.
My work spans factory safety, industrial time studies, wearable health AI, creative tools, and robotics. I like the whole process: understanding a problem, building the system, and seeing it work in someone’s hands.
A privacy-first app that turns “I wish I had someone to do this with” into a small, real meetup. AI does the awkward coordination; humans approve every step.
What we did
Built an Instagram-style meetup planner: describe a plan in plain words, get matched with compatible people, vote on real nearby places, then chat.
How we built it
A gradient-boosted matcher trained on 300k real Meetup.com join decisions picks people, Meta’s Muse models handle intake, venues and photos, and a FastAPI policy layer enforces every privacy rule.
Real-time PPE detection, from model training to a live edge deployment at Texas State.
What we did
Built real-time PPE detection and put it into daily use at Texas State’s Ingram School of Engineering, with 25–30 ms inference.
How we built it
A CBAM-enhanced YOLOv8 model fine-tuned on three PPE classes, validated on 18,000+ unseen images, optimized with TensorRT and deployed on a Jetson AGX Orin.
An iOS planner with Dash, an animated 3D companion who turns your sleep, calendar and goals into a day you can actually follow, and talks it through with you.
What we did
Built a native iOS app that reads Apple Health and your calendar, estimates your energy, schedules your goals around real events, and holds spoken mentor conversations.
How we built it
SwiftUI with HealthKit and EventKit, a Blender-modelled Dash running in RealityKit, a small energy model trained on the PMData dataset, Gemini Live for voice, and RevenueCat for Pro.
A voice-enabled 3D digital twin of your body. Your watch tells you what happened; BioTwin tells you what it means, forecasts your recovery, and helps plan your day around it.
What we did
Built a talking health twin: import wearable data, see it on an animated 3D avatar and live charts, ask questions out loud, and book recovery or workout time on your calendar.
How we built it
React + Three.js front end, FastAPI + MATLAB modelling back end, Gemini answers grounded in computed context and validated before display, and ElevenLabs speech synced to the avatar.
A table robot arm that sees people, greets them with gestures and a voice, listens, talks back, and plays hand-controlled arcade games, all running locally on an NVIDIA Jetson.
What we did
Turned a ROBOTIS OpenMANIPULATOR-X arm into the Ingram Hall Makerspace’s greeter: it waves at visitors, holds spoken conversations, and runs three games controlled by your hand.
How we built it
ROS 2 Humble on a Jetson Orin NX, YOLO and YOLO-World for vision, Whisper for hearing, a local Llama 3.2 3B for thinking, Piper for speech, and MediaPipe hand tracking for the games.
ROS 2Jetson Orin NXYOLO-WorldWhisperOllamaMediaPipe
A chess-playing robot arm, hanging upside down over a real board. A camera watches, Stockfish picks the moves, and the arm moves the pieces.
What we did
Built the chess core, the touch-screen game and the arm motion for a robot that plays you, or itself, on a real wooden board.
How we built it
Stockfish with four difficulty levels and python-chess, a Tkinter game screen, and custom kinematics and safety checks that drive an OpenMANIPULATOR-X through ROS 2 on a Jetson Orin NX.
Research, product thinking, and the engineering that brings it all together.
Sep 2024 — Present
HUMAN DIGITAL TWIN & COMPUTER VISION
Undergraduate Research Assistant
Texas State University · Ingram School of Engineering
Built a real-time MediaPipe pipeline to identify hazardous factory actions and trigger voice alerts for visually impaired workers; deployed at a San Antonio facility.
Optimized CBAM-enhanced YOLOv8 for PPE detection on Jetson AGX Orin with TensorRT.
Improved action recognition accuracy by 10% on industrial footage with YOLOv8, MediaPipe, and a TimeSformer trained on Assembly101.
Research also includes worker fatigue detection and automated MOST analysis.
Sep 2024 — Present
COMPUTER VISION STARTUP
Tech Lead · MOSTify
BobCatalyst Innovation Accelerator · Texas State University
Led the computer vision pipeline from dataset collection through training, optimization, and deployment.
Reduced 3–5 days of manual time-study analysis to 1–2 hours of semi-automated work.
Conducted 40+ customer discovery interviews; supported by a $5,000 BIAP II grant.
2024
TEACHING & ACADEMIC SUPPORT
Student Assistant
Computer Science Department · Texas State University
Supported students in C++ programming, data structures, and object-oriented programming.
Proctored exams, graded assignments, and led review sessions.
Apr 2023 — Dec 2023
FULL-STACK ENGINEERING
Software Development Intern
Orgware Construct Pvt. Ltd. · Nepal
Built a Flask and React support platform connecting rural businesses, schools, and offices with IT teams; average issue resolution took under 20 minutes.
Redesigned the MySQL schema and slow queries to support 100 concurrent users without performance degradation.
Wrote backend unit and integration tests, collaborated through pull requests, and helped deploy and monitor the application on Linux.
04 THE TOOLKIT
A full-system perspective.
The right tools for the problem. From a model to the interface around it.
