Computer Vision & Deep Learning Engineer

About the Role

We are looking for a Computer Vision & Deep Learning Engineer to design and optimize AI models for our edge-based computer vision systems. You will work on video inference pipelines, model deployment, and hardware acceleration, ensuring that our edge devices deliver real-time performance under constrained environments. This role requires expertise in deep learning frameworks, computer vision algorithms, and model optimization techniques for deployment on edge AI accelerators.

Responsibilities

• Develop and optimize computer vision and deep learning models for real-time video analytics.
• Build video inference pipelines (frame extraction, pre/post-processing, batching, asynchronous execution).
• Prepare and manage video datasets using tools like CVAT, LabelImg, and custom annotation workflows.
• Train, fine-tune, and evaluate models for object detection, classification, segmentation, and tracking (YOLO, Faster R-CNN, DeepSORT, etc).
• Apply model optimization techniques (quantization, pruning, TensorRT, OpenVINO) for performance on constrained edge devices.
• Deploy and benchmark models on AI accelerators (NVIDIA Jetson, Intel Movidius, Hailo, Coral TPU, etc).
• Collaborate with the C++ edge team to integrate AI models into production video pipelines.
• Monitor and improve accuracy, performance, and resource utilization of deployed models.
• Research and prototype new CV/AI approaches for multimodal (video + audio) analytics and transformer-based architectures.

Requirements

• 4+ years in computer vision and deep learning development.
• Proficiency with PyTorch, TensorFlow, and ONNX.
• Hands-on experience with video inference pipelines (frame extraction, pre-processing, batching).
• Strong knowledge of model quantization and acceleration (INT8/FP16, TensorRT, OpenVINO).
• Experience preparing and annotating video datasets with tools like CVAT, LabelImg.
• Familiarity with object detection, classification, segmentation, and tracking models (YOLO, Faster R-CNN, DeepSORT, etc).
• Background in AI hardware acceleration for edge devices (Jetson, Movidius, Hailo, etc).

Nice to Have

• Experience with multimodal AI (video + audio fusion).
• Knowledge of transformer-based models for vision and video understanding.
• Familiarity with real-time constraints in embedded/edge systems.
• Exposure to MLOps tools for dataset versioning, model training pipelines, and deployment automation.
• Contributions to open-source CV/AI projects.