EVM builds camera-based navigation for autonomous systems: working out where a vehicle is from what its camera sees, when satellite positioning is weak or missing. It's led by Jonathan Thursby, who spent 25 years building cameras for oil and gas wells at EV Offshore, and Matthew Gibson-Ford, our technical lead.
What you'll do
- Build models that match live camera frames to reference imagery and give a position fix.
- Decide which parts of the image to trust (learned evidence selection), and when to ignore the camera.
- Make the system report how confident it is, and recover a fix after it has lost one.
- Plan the tests, and show results on real flight data, not just on benchmarks.
- Help hire and guide two more ML engineers.
You'll have
- An MSc or PhD in computer vision, ML or robotics, or equivalent experience.
- Hands-on experience of localisation, place recognition, image matching or SLAM.
- Strong Python and PyTorch. C++ is a plus.
- Experience getting models running on small edge hardware (TensorRT, ONNX, Jetson).
Nice to have: aerial or satellite imagery, simulation and synthetic data, publications.
Pay: around £70,000–£80,000 plus share options, depending on experience. Full time, Norwich.
This role depends on project funding we've applied for. We expect to confirm by early 2027, with a start date of 1 April 2027.
Employer: Extreme Video Limited, working on EVM. Based in Norwich.
#J-18808-Ljbffr
