International recognition for JØ7 Vireo in computer science.
I build intelligence that leaves the lab.
I’m Jivesh Ramnath — a computer science and statistics student engineering machine-learning systems that have to survive real hardware, real constraints and real users.
BUILDTEST
DEPLOY
Current trajectory: on-device QNN perception, autonomous rover navigation, crash-risk modelling and the next generation of JØ7 Vireo.
Scroll to navigateProjects built around consequence, not decoration.
The common thread is deployment: constrained compute, uncertain environments, limited budgets and measurable outcomes.
J7 VireoAssistive vision / flagship system
Offline wearable vision using object detection, distance awareness, voice feedback and haptics.
02RoverCam QNNOn-device vision / Snapdragon
An eight-class YOLO detector compiled through Qualcomm QNN and deployed on the Galaxy S24 Ultra NPU.
03Mars RoverAutonomy / mechanical systems
Six-wheel Ackermann rover, simulated perception and target-seeking autonomy under competition constraints.
04Road Risk AtlasData science / public safety
A South African crash-risk model designed to estimate high-risk hours and road segments.
One system, many layers.
I work across the full path from dataset to decision, then into the physical world.
Technical stack
The work has been tested outside my laptop.
Competition results are not the goal. They are useful evidence that the engineering holds up under scrutiny.
Regional gold and top innovation recognition.
Overall African win with repeat Best Design recognition.
Cybersecurity and farm-security systems built under time pressure.