Computer Vision Engineer

What you'll do

Set the technical bar across our vehicle‑detection and license‑plate‑recognition stack.

• Govern model architecture choices, dataset strategy, evaluation pipelines, and deployment patterns.
• Design, train, optimise, and deploy CV models for vehicle detection, classification, type, colour, brand, model, and GCC license‑plate recognition.
• Convert and optimise models (YOLO family and others) for inference on Intel CPUs using OpenVINO; profile and reduce latency on edge hardware until you hit the SLA you own.
• Drive performance work on inference latency, model footprint, and CPU resource use for fleet‑scale deployment.
• Govern the labelling guideline with the AI labelling operator, prioritise edge cases, audit dataset quality, and sign off on dataset releases.
• Partner with the full‑stack and web teams to expose model outputs through stable APIs that survive model updates.
• Mentor the AI labelling operator and any junior engineers who join later.
• Review code and model evaluations.
• Stay current on practical CV research, bring back what's worth integrating, reject what isn't.

• 5+ years of production computer‑vision work, including at least one role with senior or technical‑lead responsibility.
• Deep C++ (C++17 or newer) and Python; comfortable in both Linux and Windows build environments.
• Strong OpenVINO experience or equivalent (Tensor
  
  RT, ONNX Runtime). You have personally optimised a model below an SLA you owned.
• ANPR / OCR / vehicle‑recognition or comparable safety‑critical CV background – ideally on edge hardware.
• Demonstrated ability to design model evaluation pipelines and ship measurable improvements over baselines.
• Linux command‑line fluency.
• English fluency – written and spoken.
• Nice to have:
  Direct GCC license‑plate experience, experience working with classified or restricted‑access datasets and the operational disciplines that go with them, experience leading 1‑3 person AI teams, C++ inference on Intel platforms (OpenVINO model optimisation, INT8 quantisation, multi‑threaded inference), Windows desktop application packaging, experience with M1 / Apple Silicon for future portability, Arabic reading useful for GCC plate work.

Abu Dhabi – onsite at our headquarters. Open to candidates already in the UAE with valid work authorization, or to candidates willing to relocate; the company sponsors residency for international hires. Hybrid available after probation by manager agreement.