وظيفة مهندس خوارزميات الرؤية لدى Solar-LIT في جدة

I. Vision Perception Algorithm Development

Responsible for the vision perception module of PV (photovoltaic) cleaning robots, including but not limited to: object detection and instance segmentation of PV modules, navigation path recognition, detection and classification of obstacles (such as cables, posts, and weeds), and analysis of surface contaminants on modules (dust, bird droppings, snow accumulation).

Responsible for the full pipeline of vision models in industrial scenarios, including data collection, cleaning, annotation scheme design, model selection, training, optimization, and deployment.

II. Model Deployment & Tuning

Deploy advanced 2D/3D vision algorithms and deep learning models (e.g., the YOLO series, Mask R-CNN, Segment Anything Model, etc.) onto robotic embedded platforms (e.g., the Jetson series), and perform performance tuning to balance speed and accuracy.

III. Exploration of New Technologies

Explore fusion strategies combining visual information with sensor data such as LiDAR and IMU, to provide more robust perception inputs for robot localization, navigation, and decision-making.

Track the latest advances in computer vision and deep learning (e.g., Transformers, Vision-Language Models, domain adaptation, etc.), and evaluate their application potential for improving the robot's environmental understanding and level of intelligence.

IV. Solution Design & On-Site Debugging

Collaborate closely with the algorithm development, hardware, and mechanical teams; participate in requirement reviews and technical solution discussions.

Participate in on-site debugging to address vision challenges caused by complex lighting (strong light, reflections, low light), seasonal changes, and weather conditions (rain, fog), continuously improving the algorithm's environmental adaptability.

Master's degree or above in Computer Science, Artificial Intelligence, Automation, Electronic Engineering, or a related field.

2+ years of work experience, with a deep understanding of the fundamental principles of computer vision and deep learning, including image processing and architectures such as CNN, RNN, and Transformer.

Proficient in tasks such as object detection and instance/semantic segmentation, with extensive hands-on experience training and tuning relevant models (e.g., Faster R-CNN, YOLO, Mask R-CNN, DeepLab, etc.).

Strong sense of responsibility, a team-oriented spirit, and excellent communication skills.