Senior Real-Time Perception Engineer - Industrial AI

Cognistic; Kentucky; On-site; USD 110,000 - 180,000; Full time

Job summary

Cognistic is seeking a Senior Perception Engineer to design and deploy real-time perception systems for industrial and autonomous applications. You will work on 2D/3D perception, sensor fusion, and deep learning integration, delivering production-ready C++ software with GPU acceleration on NVIDIA platforms.

The role requires hands-on experience with vision algorithms, SLAM-era techniques, and deployment workflows using TensorRT and ONNX.

Qualifications

• Bachelor's or Master's degree in Computer Science, Computer Engineering, Robotics, Electrical Engineering, AI, or a related field.
• Strong professional experience in Computer Vision, Robotics, Autonomous Vehicles, ADAS, or similar perception systems.
• Strong knowledge of modern computer vision, 3D perception, and deep learning.
• Hands-on experience with Object Detection, Semantic Segmentation and Instance Segmentation, Image processing and computer vision.
• Experience working with 2D cameras and at least one type of 3D sensor such as Stereo/3D cameras, LiDAR or Depth cameras.
• Strong programming skills in C++.
• Good knowledge of Python for development, experimentation, and model evaluation.
• Strong understanding of software optimization and real-time system performance.
• Experience deploying and optimizing deep learning models using TensorRT.
• Experience with ONNX model conversion and deployment workflows.
• Experience with GStreamer and developing real-time video processing pipelines.
• Experience with NVIDIA DeepStream SDK and GPU-accelerated video analytics.
• Experience developing or deploying applications on NVIDIA Jetson platforms or NVIDIA GPUs.
• Good understanding of Linux development environments.
• Experience with Git and modern software development practices.
• Strong debugging, analytical, and problem-solving skills.

Responsibilities

• Design, develop, and maintain real-time perception components for industrial and robotics systems.
• Develop and integrate object detection, semantic/instance segmentation, tracking, and other vision-based perception algorithms.
• Work with sensors including 2D RGB, Stereo 3D cameras and LiDAR.
• Process and analyze 2D image data and 3D point cloud/depth data.
• Develop and optimize algorithms for 3D point cloud processing: filtering, segmentation, clustering, obstacle detection, geometric analysis.
• Integrate deep learning models into production C++ perception pipelines.
• Optimize perception models for real-time inference on NVIDIA Jetson and GPUs.
• Convert, deploy, and optimize neural networks using ONNX and TensorRT.
• Develop and maintain real-time camera and video processing pipelines using GStreamer and DeepStream.
• Profile and optimize CPU/GPU performance, memory usage, latency, and throughput.
• Develop and maintain robust, production-quality C++ software for embedded systems.
• Integrate cameras and sensors into real-time processing pipelines.
• Debug and troubleshoot perception, sensor, hardware, and performance issues on deployed systems.
• Participate in system testing, field testing, calibration, validation, and deployment.
• Contribute to perception architecture and technical design decisions.

Skills

Computer Vision

Robotics

Autonomous Vehicles

ADAS

Real-time systems

Deep learning deployment

C++ programming

Python programming

Linux development

Git version control

Education

Bachelor's or Master's degree in CS/CE/Robotics/EE/AI

Tools

GStreamer

NVIDIA DeepStream SDK

TensorRT

ONNX

Jetson platform

NVIDIA GPUs

Git

Linux

Job description

Cognistic is seeking a Senior Perception Engineer to design and deploy real-time perception systems for industrial and autonomous applications. You will work on 2D/3D perception, sensor fusion, and deep learning integration, delivering production-ready C++ software with GPU acceleration on NVIDIA platforms.

The role requires hands-on experience with vision algorithms, SLAM-era techniques, and deployment workflows using TensorRT and ONNX.