Computer Vision Engineer

Computer Vision Engineer — Autonomy (Remote)

We build retrofit autonomy modules for existing UAV fleets operating in GPS-denied environments. This is real-world deployment, not research or simulation. Constrained hardware, degraded comms, systems that have to work first time.

Kessari is moving from TRL 6 to TRL 8 with active partners.

We need someone to own perception through to targeting, end-to-end.

What you’ll do

• Build and deploy real-time detection and tracking pipelines on edge hardware
• Take models from training through optimisation into field deployment
• Work on low-latency systems running on constrained GPUs (Jetson class)
• Handle messy real-world data (aerial, oblique, thermal, small objects)
• Ship systems that run at 30+ FPS in production
• Work in GPS-denied conditions where localisation and perception must hold up under uncertainty

You’re a fit if you can

• Train and deploy object detection models (YOLO, RT-DETR or similar)
• Optimise models for real-time edge inference (TensorRT, ONNX or similar)
• Implement multi-object tracking (ByteTrack, BoT-SORT or similar)
• Own the data pipeline (collection, annotation, validation)
• Work across Python and some C++, Linux, Docker
• Strong bonus
• Experience in GPS-denied navigation or perception systems
• Drone or aerial imagery experience
• Thermal or infrared perception
• Visual SLAM or odometry integration
• CUDA or GPU optimisation
• Synthetic data or simulation

What matters

This is not a research role. You need to ship fast, handle ambiguity, and make systems work in the field.

Comp

€90k – €140k+ depending on level

Equity and performance upside tied to deployments

• DM directly to apply