Andre Mossi Experience
Computer Vision System 2024 To 2025
In a team of three, we engineered an industrial traceability system to automatically detect and log pallet movements inside a production facility. The previous workflow relied on manual inspection of damaged pallets, resulting in delays and human error. The system used dual cameras to detect motion and generate event-based records containing captured images and object detections. These events were processed using Ultralytics YOLO running on NVIDIA Jetson (migrated from Arduino due to overheating and computational limitations) and stored in TimescaleDB. A custom web platform provided real-time visualization and system control.
Engineering Decisions & Ownership
- Migrated hardware from Arduino to NVIDIA Jetson to resolve thermal and performance constraints.
- Containerized backend services using Docker for reliable deployment.
- Designed event-based database modeling for scalable storage and retrieval.
- Developed 100% of the web interface and contributed ~50% across backend, hardware integration, and AI deployment.
- Implemented image deletion, detection editing, and retraining pipeline integration.
Operational & Business Impact
- Conducted bi-weekly progress demonstrations to secondary management.
- Presented milestone reviews to executive leadership at mid-project and final phases.
- Secured continued funding based on measurable progress and system validation.
- Project continuation depended on performance delivery; failure to meet milestones would have resulted in termination.
- TypeEducation
- ContentArtificial Intelligence, Data Entry, Data Analysis, JavaScript, React, C++, Shell, Docker, Python, Rust, Ultralytics Yolo, Machine Learning, TimescaleDB, Data Modeling
