The client's video never leaves their site. You make vision models run on the hardware that is there: with little memory, no big GPU and no reliable internet.
The role
- Optimize detection and tracking models for edge hardware.
- Choose and test hardware for each type of site.
- Make the equipment work without internet and recover on its own.
- Measure real performance on site, not in the lab.
Your first 90 days
- Day 30
You run the vision pipeline on an edge device and measure it.
- Day 60
You deliver an optimized version that runs in real time.
- Day 90
A client site runs on your configuration.
Stack and tools
PythonC / C++ONNXOpenCVLinuxffmpeg
What we look for
Must have
- Python and some C or C++.
- You have worked with hardware: boards, cameras or sensors.
- You know how to measure performance and memory.
- Going on site to fix things doesn't scare you.
Nice to have
- Model quantization and optimization.
- Embedded Linux.
- RTSP and ONVIF.
What working here is like
What you build goes into products that operate real organizations. You work directly with the founders, with no layers between your code and production, under the same rule we give the intelligence: a person signs what is critical.