is an early-stage technology company building retail technology with computer vision. We are hiring the engineer who will own our vision system and lead the vision team.
What we want
- Four or more years in computer vision
, at least three with your models running in production on real cameras.
- Shipped object detection, people detection, multi-object tracking and person re-identification
, and can explain what broke and how you fixed it.
- Deployed quantised models on edge hardware
(Jetson, Intel iGPU or similar) with TensorRT, OpenVINO or ONNX Runtime.
- Built or run
GStreamer, DeepStream or FFmpeg pipelines against RTSP cameras.
- Hands-on with vision-language models
: run them locally, ideally fine-tuned or distilled one.
- PyTorch, robust Python, working C++, comfortable on Linux.
- You measure precision, recall, latency and cost per stream,
and can state a false-positive rate before and after a fix
.
- Ready to lead a small team and write architecture others can build against.
What you will do
- Design and build the vision system: real-time detection, tracking and re-identification on live camera feeds.
- Bring vision-language models into production where they earn their place.
- Run inference on edge hardware with privacy masking applied on the device.
- Keep false positives low and prove it with numbers.
- Hire and lead the vision team after funding.
Compensation:
- Until funding:
a basic salary plus ESOP.
- After funding:
salary normalised to market, ESOP continues.
NO RESUME : fill the form : https://forms.gle/1t797YTkGqy33exZ6