Mobile & IoT
A complete embedded-ML IoT safety system, firmware to dashboard
Role
Developer
Duration
One Trimester
Date
2025
Type
University Project
4
System Layers
Firmware → Web
Stack Span
On-device
Inference
A four-layer system built end to end. An Arduino-based wearable written in C++ runs an Edge Impulse-trained fall-detection model on-device through EloquentTinyML, streaming from accelerometer and gyroscope sensors; a Node.js API with JWT auth, Joi validation, helmet, and rate limiting ingests events and triggers alerts; and a Next.js dashboard gives caregivers live monitoring. One of the few student projects where hardware, embedded ML, and web engineering all had to work together.
What it does, how it works, and what it ships.
Detecting a fall reliably from noisy accelerometer data, while keeping the pipeline fast enough that an alert reaches a caregiver when it matters.
Raw accelerometer-gyroscope data trains an Edge Impulse model that is exported to the wearable for on-device inference, so there is no cloud round-trip for the classification decision. The raw sensor dataset ships in the repo.
Hardware-to-cloud engineering in one project.
C++
Wearable Firmware
TinyML
On-Device Classifier
Live
Alert Dashboard
“IoT projects punish hand-waving. Every layer either works or the demo fails in front of everyone.”
Reach out for source code walkthroughs or collaboration.