Projects
Here are some of my projects showcasing my skills and expertise
Hybrid Exoplanet Recognition and Model Evaluation System - HERMES
HERMES is a web application designed for the NASA Space Apps Challenge 2025, helping users classify exoplanet candidates efficiently. It allows users to upload or edit exoplanet datasets and run a machine learning model to identify confirmed planets, potential candidates, or false positives.
Features:
- Exoplanet Classification Engine
- Dataset Upload & Editing
- Interactive Visual Analytics
- Manual Model Retraining
- Web-Based Interface
Smart Packaging Control & Demand Forecasting Web Application
Developed a web application deployed on a Raspberry Pi to remotely monitor and control a packaging machine. Integrated a SARIMA forecasting model into the app to predict daily packaging demand, achieving a 4.61% MAPE. The project scope focused solely on the software and system integration.
Features:
- Automated packaging process with real-time monitoring and control
- Intelligent demand forecasting using SARIMA with monthly retraining
- IoT-enabled Raspberry Pi integration for seamless machine operation
- Data-driven optimization of packaging workflow and production tracking
AI-Powered Mosquito Monitoring System
Research and prototype of an attractant-based trap with image recognition to identify Aedes mosquitoes, with a companion mobile app for real-time location monitoring and notifications. The system's detection accuracy was evaluated under engineering performance and efficiency constraints.
Features:
- Deep learning-based Aedes detection
- Integrated image capture and classification
- GPS geotagging of trap location
- Mobile app alerts and monitoring
Automated Remote Geospatial User Service - ARGUS
ARGUS is a Web platform for NASA Space Apps Challenge 2024, that allows users to view multispectral satellite images and check future satellite pass schedules by entering latitude and longitude coordinates, providing interactive map-based visualizations and scheduling information via APIs.
Features:
- Interactive map-based display of multispectral satellite imagery
- Satellite pass prediction using real-time API data
- User-friendly interface for entering coordinates and retrieving data
- Integration of imagery and scheduling information for easy planning
Exercise Posture Suggestion System
An Exercise Posture Suggestion System using deep learning to analyze and provide real-time feedback on correct exercise posture. Integrated pose estimation, human activity recognition, and error detection algorithms to offer personalized guidance. The software interface, built in Unity, allows users to visualize correct postures and optimize workouts, aiming to reduce injury risk and improve exercise effectiveness.
Features:
- Real-time exercise posture analysis with deep learning
- Pose estimation using ResNet-50, YOLOv8, and YOLO-NAS
- Human activity recognition with Conv(2+1)D, 3D CNN, and CNN-BiLSTM
- Interactive Unity interface for feedback and posture visualization
COVID-19 Symptom-Based Quarantine Prediction
A Research about Machine learning models using Israeli Ministry of Health data to predict individuals requiring quarantine based on common COVID-19 symptoms, offering a faster, accessible alternative to testing while reducing false positives.
Features:
- Symptom-based risk prediction with supervised ML models
- Data preprocessing and feature engineering for improved accuracy
- Real-time quarantine recommendations for rapid intervention
- Performance evaluation to reduce false positives