Flagship project
JØ7 VIREO — Assistive vision platform
A wearable offline assistive system that uses a custom-trained YOLOv8m model to detect 100 everyday object classes,
estimate distance, and provide audio feedback to visually impaired users.
- Runs on edge hardware without cloud dependence.
- Combines camera vision, distance awareness, and haptic safety feedback.
- Built around low-cost hardware and reuse of repurposed lithium-ion batteries.
- Designed for real-world navigation, privacy, and accessibility.
YOLOv8m
Raspberry Pi
Hailo
LiDAR
Audio feedback
Competition engineering
Cars4Mars African Rover Challenge
A rover systems project built through the Cars4Mars competition, combining 3D CAD, autonomous thinking, mechanical design,
electronics, and programming under contest constraints.
- Team Martian Mechanics won Best Design Award two years in a row.
- Placed first this year and won the competition overall.
- Demonstrates mechanical design, systems thinking, and reliable build execution.
- Useful proof of practical engineering ability under pressure.
3D CAD
Robotics
Systems design
Teamwork
ML pipeline work
Dataset annotation, training, and deployment
Work focused on the complete machine-learning lifecycle: collecting data, annotating it properly, training models,
evaluating performance, and deploying on constrained devices.
- Hands-on experience with data annotation workflows.
- Model training and debugging in PyTorch.
- Performance-focused deployment for edge hardware.
- Built around speed, accuracy, and reliability tradeoffs.
PyTorch
OpenCV
Annotation
Deployment
Technical growth area
Embedded AI and sensor fusion exploration
Ongoing work connecting sensors, control systems, and computer vision so that software decisions map to physical-world actions.
- Raspberry Pi and ESP32 integration.
- Arduino IoT programming and hardware interfacing.
- PCB design and electronics prototyping.
- Practical focus on portability and cost.
ESP32
IoT
PCB design
Sensors