Projects

Parkinson's Care VR System
github slides post
C# Unity TensorFlow React Flask SVM CNN
Georgetown University H2AI Hackathon: 1st Place Grand Prize & Patient Safety Award
  • Architected VR-based medical assessment tool for Parkinson's patients, integrating Unity with TensorFlow ML models to process four concurrent biometric streams for real-time symptom quantification with 92% clinical accuracy.
  • Designed event-driven architecture for doctor-patient remote monitoring, connecting VR sensors to Python backend via WebSocket protocol to enable sub-100ms latency for medical data visualization.
WALL-E
github devpost post
Python React GPT-4 API PyQT5 OpenCV C++ Embedded Systems
PatriotHacks 2024: Triple Winner (Patriot Favorite, Most Likely to be a Startup, Best Cyberpunk Theme)
  • Engineered dual-purpose cleanup robot and file organizer, combining autonomous waste detection with LLM-powered digital file management to achieve 95% classification accuracy through custom prompt engineering.
  • Implemented hardware abstraction layer for physical robot navigation, bridging Python computer vision models with C++ microcontroller firmware for real-time object detection on embedded systems.
Peekabot
github devpost post
Python OpenCV AWS IoT Mediapipe C++ Raspberry Pi IoT Systems
HackOverflow 2024: Best Robot Hack Winner
  • Built IoT child safety monitoring robot using Raspberry Pi computer vision and Arduino motor control, tracking toddler movements in real-time to detect dangerous situations and send instant alerts.
  • Architected home safety system integrating OpenCV/Mediapipe pose detection with AWS IoT cloud alerts and ngrok live streaming, enabling parents to monitor and receive notifications remotely.
NaviguideAI
github devpost post
Java Python TensorFlow ResNet Computer Vision PyAudio ESP32
PatriotHacks 2023: Best AI-Powered Hack Winner
  • Developed accessibility navigation aid for visually impaired users, using CNN-ViT fusion architecture with ResNet-18 backbone to achieve 92% obstacle detection accuracy at 127ms inference time.
  • Engineered spatial audio guidance system for blind navigation, integrating PyAudio 3D sound synthesis with custom haptic feedback to enable 96% successful navigation through complex environments.
Optimal Path Navigation
github
Python TensorFlow Hugging Face 3D Mapping Depth Estimation A* Algorithm
  • Built indoor navigation system for visually impaired users, converting 2D images to 3D point clouds using Hugging Face transformers and solving camera calibration challenges for real-time environmental mapping.
  • Implemented safe path planning for obstacle avoidance, using custom A* algorithm on projected 2D occupancy grids to enable dynamic navigation with sub-5 second route computation.