During a hackathon, the pressure is on as soon as the countdown starts... Students learn about various challenges, technologies, understanding their teammates' strengths and brainstorming their project. For team Rocko, Mohammad Steitieh, Amr Altaweel, Said Elakad, Bader El-Asadi, sometimes the best ideas come from the most unsuspecting places.

Inspiration struck after Amr watched a video about the Nutty Putty Cave disaster. He pitched the idea the next morning, the team was hooked and roughly 32 hours later, they'd built a working emergency beacon that won first place in the QNX hardware track at cuHacking 7, beating out 20 competing projects.

“In line with the QNX ethos, we wanted our project to provide a real solution to a mission-critical problem. Amr proposed the cave diving safety paradigm, we had quantum magnetometers on hand due to my work with the Entrust Securities Lab at Carleton University, magnetic fields cut through rock, we knew we had to make this happen.” - Mohammad Steitieh

Cave environments are unforgiving for communication. Radio frequency (RF) signals weaken severely through rock and soil, so a trapped or injured explorer deep underground often has no viable path to call for help.

original

The key difference? Send the diagnosis, not the data. By classifying the emergency on-device and encoding it into just two bytes, minimizing the payload that has to survive one of the harshest transmission channels imaginable, in emergency situations.

The team ran two entirely local AI architectures on a QNX-equipped Raspberry Pi 5:

Speech-to-text via OpenAI Whisper to capture the spoken emergency.
Injury classification using a locally trained MobileNetV2 CNN with a custom softmax layer. It takes a 224×224 wound image and predicts injury class, exported to TFLite to run efficiently on the Pi.

For the physical layer, they hand-coiled 20 meters of 18-gauge copper wire around a CD cylinder to form a magnetic transmitter. The 12-bit message uses Manchester encoding at 8Hz to guarantee bit distinguishability, with a tilde delimiter marking the start of each frame. On the surface, a quantum TMR magnetometer with a 7–9Hz bandpass filter and matched filtering recover the signal from the noise floor. Embedded AI and quantum sensing, working together and orchestrated on QNX OS for reliability.

“I remember staying with Said late into the night tinkering with the QNX Raspberry Pi. In our world, heavy agentic workflows are everything. It was a change of pace to work with something lightweight by design. The architecture of QNX is its crowning feature; it's built for scenarios where it can never fail. For us, this was a new challenge with a quantifiable return, we are engineers, we love that.” - Bader El-Asadi

That's why hands-on access to industrial-grade platforms matters. Through QNX Everywhere, students can prototype on the same foundational RTOS used across automotive, medical, and other safety-critical industries, for free, on hardware as accessible as a Raspberry Pi.

Rocko is proof that with the right inspiration, and the right tools, students can build technology that genuinely saves lives. Congratulations to Team Rocko on such a successful weekend, we can't wait to see what they do next.