Engineer - Perception (Deep Learning) (m/f/d)

Halian; Abu Dhabi; On-site; AED 150,000 - 200,000; Full time

Job summary

Halian is seeking a Perception Engineer in Abu Dhabi to develop and enhance a perception stack utilizing both classical computer vision techniques and modern deep learning approaches. The ideal candidate will work on end-to-end learning systems for autonomous navigation across diverse environments.

Responsibilities include designing and implementing real-time perception modules, coding in C++ and Python, and integrating various data sources for improved object reasoning. Candidates must have a strong foundation in deep learning and real-time systems, alongside excellent communication skills.

Qualifications

• 4+ years of experience in real-time perception or computer vision systems.
• Strong understanding of deep learning architectures and their application in perception tasks.
• Experience with ARM64-based edge devices.

Responsibilities

• Design and implement robust, real-time perception modules.
• Develop clean, modular, and efficient code in C++ and Python.
• Conduct field testing and system tuning for performance.

Skills

Real-time perception

Computer vision

C++ programming

Python programming

Deep learning

OpenCV

PCL

PyTorch

TensorRT

Education

Bachelor’s or Master’s degree in Robotics, Computer Vision, or related field

Tools

ROS1/ROS2

Docker

Git

CI/CD systems

Job description

Role Overview

Seeking a Perception Engineer to develop and enhance a perception stack using both classical computer vision techniques and modern deep learning approaches. The role focuses on end-to-end learning systems while maintaining strong grounding in traditional methods. The work supports real-time, deployable perception systems for autonomous navigation across diverse environments and robotic platforms.

Key Responsibilities

• Design and implement robust, real-time perception modules for both structured (on-road) and unstructured (off-road) environments using learning-based and geometric methods.
• Develop clean, modular, and efficient code in C++ and Python, following ROS1/ROS2 and industry best practices.
• Conduct extensive field testing and iterative system tuning to ensure performance and reliability under real-world conditions.
• Work closely with adjacent system components such as localization, planning, and control to enable seamless integration.
• Participate in code reviews and maintain clear, detailed technical documentation.
• Integrate HD map data (e.g., Lanelet2 and semantic layers) into perception pipelines to support object-level reasoning, localization, and terrain adaptation.

Minimum Qualifications and Skills

• Bachelor’s or Master’s degree in Robotics, Computer Vision, or a related field.
• 4+ years of experience in real-time perception or computer vision systems.
• Proficiency with libraries and tools such as OpenCV, PCL, Open3D, PyTorch, and TensorRT.
• Strong understanding of deep learning architectures (e.g., CNNs, transformers) and their application in perception tasks.
• Familiarity with modern perception approaches such as LSS, BEVFormer, BEVFusion, and ability to assess trade-offs between accuracy, latency, and deployability.
• Strong programming skills in Python and C++, with experience in ROS1 or ROS2 systems.
• Experience with modern development workflows and tools such as Git, Docker, and CI/CD systems.

Preferred Qualifications and Skills

• Experience combining traditional perception techniques with deep learning-based methods.
• Familiarity with experiment management and MLOps tools such as Hydra, Weights & Biases, AWS SageMaker, or GCP.
• Knowledge of HD map frameworks such as Lanelet2 or similar systems.
• Experience working with ARM64-based edge devices, including NVIDIA Jetson platforms.

Additional Skills

• Strong communication skills with the ability to explain complex technical concepts clearly.
• Ability to work independently in fast-paced environments.
• Comfortable balancing experimental prototyping with debugging real-world system challenges.