Leadership
Adam Hassan
Team Lead & Head of Marketing
Project Contributions
- Led the team across all project phases, coordinating between hardware, software, and ML workstreams to deliver a fully integrated system
- Defined the overall project scope, timeline, and deliverables for the Recycla smart bin system
- Developed the go-to-market strategy and brand positioning for presentation to stakeholders and the university
- Managed all stakeholder communications, progress reporting, and milestone tracking throughout the development cycle
- Oversaw the creation of all marketing materials, pitch decks, and the promotional website content
- Drove cross-functional integration to ensure hardware, software, and ML components worked together seamlessly into a cohesive product
Software Design & Development
Zivan Erdevicki
Head of Software Design & Development
Project Contributions
- Designed and built the complete software pipeline from image capture through waste classification to physical sorting output
- Developed the Python-based inference engine that loads and executes the TFLite model on the Raspberry Pi 4
- Implemented the GPIO control logic that translates classification results into precise servo motor actuation signals
- Built the sensor integration layer that connects the ultrasonic trigger event to the camera capture and inference sequence
- Created the user interface and LED status display system that provides real-time visual feedback on classification results
- Designed and developed this website, handling all front-end architecture, UI/UX design, and visual implementation
Machine Learning & Integration
Bassam Alghamdi
Head of Machine Learning & Integration
Project Contributions
- Selected MobileNetV2 as the base architecture and implemented transfer learning from ImageNet pre-trained weights
- Curated, cleaned, and organized the training dataset across six waste categories: cardboard, glass, metal, paper, plastic, and general trash
- Designed the data augmentation pipeline with brightness, contrast, rotation, flip, zoom, and channel shift transforms for real-world robustness
- Trained, validated, and iteratively optimized the model to achieve reliable classification accuracy across all material types
- Quantized the Keras model to TFLite INT8 format, reducing model size and enabling efficient edge inference on the Raspberry Pi 4
- Implemented confidence thresholding logic that defaults uncertain classifications to garbage, preventing recycling contamination
Hardware & Integration
Mahmoud Rabie
Head of Hardware & Integration
Project Contributions
- Designed and fabricated the complete physical smart bin enclosure with dual sorting compartments and integrated electronics bay
- Integrated the Raspberry Pi 4, Arducam 8MP camera, and HC-SR04 ultrasonic sensor into a unified, reliable hardware platform
- Wired, calibrated, and tested dual SG90 servo motors for consistent, reliable lid actuation under repeated use
- Designed the LED lighting array that provides consistent illumination for image capture regardless of ambient conditions
- Handled all circuit design, soldering, cable management, and physical assembly of the prototype from raw components
- Ensured seamless hardware-software integration, resolving GPIO conflicts, power distribution, and timing issues across the full classification pipeline