Computer Vision Scientist

Computer Vision Scientist – L&T Group AI Center of Excellence

Scope: L&T Group-wide

Experience:

8-15 years

Reports to:

Head, AI Center of Excellence

Help L&T businesses see, measure, and understand the physical world  will apply state-of-the-art 2D and 3D vision to high-impact problems such as quality inspection, construction progress monitoring, worker safety, and asset health, and own the CoE's reusable vision platform for Engineering, Manufacturing, Infrastructure, Energy, and Defence.

• Design, train, and optimize models for detection, segmentation, tracking, pose estimation, and anomaly detection, including adapting foundation models (SAM, DINOv2, YOLO, ViT) to industrial and site imagery.
• Build 3D vision capabilities: depth estimation, point-cloud processing, photogrammetry, and scene reconstruction.
• Take solutions from proof of concept to production with clear business-value metrics, and integrate outputs with BIM, digital twins, and enterprise systems.
• Set data acquisition and labeling standards across camera, drone, thermal, and LiDAR sources; use synthetic and self-supervised methods to reduce annotation effort.
• Optimize models for real-time edge inference (NVIDIA Jetson, industrial cameras) with quantization, pruning, and TensorRT or OpenVINO; work with MLOps on monitoring and retraining.
• Mentor scientists and engineers, contribute to the CoE repository, and publish or file patents.

• Define the CoE computer vision roadmap and reusable vision platform.
• Deliver 3-5 vision proofs of concept, at least 2 scaled to production.
• Establish a standard annotation and model evaluation framework, and publish an edge deployment reference architecture.

• Master's or PhD in Computer Science, Electrical Engineering, AI, Applied Mathematics, or a related field (PhD preferred).
• 8-15 years in computer vision and machine learning, with a track record of production deployments. Industrial, manufacturing, construction, or infrastructure experience is strongly preferred.
• Deep expertise in CNNs, vision transformers, and modern detection and segmentation architectures.
• Strong Python and C++, with PyTorch, OpenCV, and 3D libraries such as Open3D or PCL; hands-on video analytics and edge AI deployment. Publications at CVPR, ICCV, ECCV, NeurIPS, or similar venues are a plus.