We are building a real-time computer vision platform that detects safety violations across multiple CCTV streams on live-fire training ranges in day, night-IR, and thermal conditions. The system has to work. Failures have real-world consequences, not just bad metrics.
We are looking for a senior computer vision engineer to help us deliver it.
What you'll work on
- Live multi-CCTV detection across day, night-IR, and thermal conditions.
- Complex and dynamic CV solutions: weapon / object orientation under occlusion, action sequence recognition for clearing drills or SOP, multi-camera Re-ID across identical uniforms, and smoking detection
- Designing algorithms and policies that combine multiple detection frameworks to drive false positives down without losing recall on critical events.
- Continuously analysing where the system is uncertain and lifting confidence calibration and recall rates for P0 use cases.
- Scalable architecture across central GPU inference and distributed edge deployment, designed to grow as cameras and ranges are added.
- Hazard analysis, validation, and field monitoring.
- Working directly with range safety officers and operational stakeholders.
What we're looking for
- 5+ years of production Computer Vision (CV) experience.
- Live multi-CCTV detection experience in production, not lab demos or batch video processing.
- Demonstrated experience across day, night-IR, and thermal imaging conditions.
- Hands-on with NVIDIA DeepStream and TensorRT (FP16/INT8 optimisation, custom plugins where needed).
- Proven ability to combine multiple detection frameworks and modalities into a single decision pipeline, with concrete examples of driving false positives down without losing recall.
- A clear method for analysing and lifting confidence calibration and recall rates.
- Shipped CV workloads on both central GPU servers and edge nodes; can articulate the architectural tradeoffs.
- Experience contributing to CV systems that scale gracefully as cameras and sites grow.
- Re-identification (Re-ID) experience, ideally in cases with weak appearance cues (uniforms, identical gear).
- Solid practical expertise in object detection, pose estimation, and multi-object tracking.
- Experience with action recognition (SlowFast / VideoMAE / TimeSformer), multi-camera 3D geometry, or audio-visual sensor fusion.
- Strong Python; comfortable with C++ or CUDA; PyTorch and modern MLOps.
Bonus points
- Prior experience on safety-critical or mission-critical systems (defence, autonomous systems, industrial safety, perimeter security).
- Visual synthetic data generation experience (Unreal, NVIDIA Omniverse, Isaac Sim).
- Solutions that bridge the sim-to-real gap (domain randomisation, domain adaptation, progressive fine-tuning).
- Deployment experience on Jetson-class edge hardware (Orin, Xavier).
- Prior work on military, law enforcement, or weapons-handling CV applications.
- Acoustic event detection or audio-visual fusion in adversarial outdoor environments.
- Familiarity with safety engineering frameworks (FMEA, STPA, ISO 21448 SOTIF).
- Published in CVPR / ICCV / ECCV / WACV, or contributed to relevant patents.
- Direct field exposure, military service, range officer familiarisation, or similar.
Why this role
- The work is genuinely interesting. Most production CV is selling more ads or counting items on shelves. This isn't.
- Real ownership of meaningful components. You will be the technical owner of multiple use cases, not a small cog in a large org.
- Real users, real feedback. You will spend time with the people who actually use the system.
How to apply
Send your CV to [Confidential Information] with the subject line Senior CV Engineer.
Include a short note (one page max) about a CV system you shipped where failure had real consequences. We care about what you learned, not just what you built.
We respond to every qualified application within two weeks.
