Hi NVIDIA team,
I’m an automation engineer/ solutions architect exploring edge computing for a personal learning project. I’ve been working with industrial data acquisition (PLC-based systems, vibration/thermal sensors) and I’m dissatisfied with current approaches that rely on CPU-based processing in the cloud or on-site servers. They’re slow, expensive to maintain, and operationally inconvenient for remote industrial sites. If this project of mine works, I can pitch it to my company and be at the frontier of turbomachinery applying the most up-to-date edge-to-ai inferences for predictive maintanence.
What I’m trying to build:
A predictive maintenance system for rotating equipment lifecycle monitoring. The concept:
- Collect vibration and thermal data from industrial sensors at the edge
- Run inference locally (anomaly detection, equipment health scoring)
- Stream only aggregated results (not raw video or large data files) to a cloud backend
- Deploy remotely without needing an engineer on-site for every update
Why I’m reaching out to NVIDIA:
I have no prior experience with GPU-based edge platforms. I keep reading that Jetson is purpose-built for this kind of local inference workload, but I don’t know where to start in your product lineup. I need guidance on:
- Which Jetson platform fits a beginner building an industrial edge inference project
- Whether Jetson can handle continuous sensor data inference in hot industrial environments (35-45°C ambient)
- How model deployment and remote updates work in practice
- What the realistic learning curve looks like for someone coming from industrial automation (PLCs, SCADA) but new to GPU programming
I’m not asking for a product pitch — I want to understand whether the Jetson ecosystem is the right foundation for this use case before I commit to the learning path and hardware purchase.
Any guidance or pointers to relevant resources would be appreciated.
Thanks,
Abdulrahman Mohammad