AI & Computer Vision Intern at Thornton Tomasetti in New York, NY

Summary: Thornton Tomasetti is an independent organization applying engineering and scientific principles to solve global challenges. They are seeking an AI & Computer Vision Intern to support their Forensics team in labeling and curating datasets for computer vision models, assisting in data pre-processing, and collaborating with forensic engineers to enhance model performance. Responsibilities: - Support the AI/ML team in labeling and curating datasets for computer vision models, including classification, object detection, instance segmentation, and semantic segmentation - Work with 2D image datasets (e.g., photos, drone imagery, plan drawings) to draw and review labels for objects and damage types - Work with 3D point cloud models (e.g., structural frames, industrial facilities) to segment and label structural components and other features in 3D space - Assist in developing and refining labeling guidelines, class definitions, and attribute schemas to improve consistency and model performance - Perform quality assurance (QA/QC) on existing labels, identify labeling errors or ambiguities, and suggest corrections and improvements - Help the engineering team prepare example datasets, figures, and visualizations for internal presentations, reports, and publications related to AI model development - Assist in basic data pre‑processing and organization (e.g., dataset splitting, file naming, simple scripting) to support experimentation and model training workflows - Collaborate with forensic engineers to understand model objectives and failure modes, and adjust labeling strategies accordingly (e.g., edge cases, hard negatives) Required Qualifications: - Computer vision and AI: basic understanding of common model types (e.g., CNNs, U‑Net, Faster R‑CNN, YOLO) and tasks (classification, detection, segmentation) - Programming and data tools: Python and/or MATLAB for simple data manipulation, visualization, and experiment tracking - Point cloud software: CloudCompare or other point cloud viewing and manipulation software for slicing, filtering, and annotating 3D datasets - Labeling platforms: experience with image or point cloud annotation tools (e.g., Supervisely, CVAT, Labelbox, or similar) - A strong interest in applying AI and computer vision to structural and forensic engineering problems is essential Required Skills: Computer vision, Convolutional Neural Networks (CNNs), U-Net, Faster R-CNN, YOLO, Classification, Object detection, Instance segmentation, Semantic segmentation, Python, MATLAB, CloudCompare, Supervisely, CVAT, Labelbox, 3D point cloud annotation Benefits: Medical, Dental, Vision, Life, AD&D, Disability and other voluntary benefits, Flexible Spending Accounts for Medical and Childcare, Paid Time Off, Family Leave for New Parents, Volunteer Time, Tuition Reimbursement, Commuter Transit (where available), 401k retirement savings with Company matching on employee contributions and/or qualified student loan repayments, Fitness Reimbursement, And other various wellness, diversity/inclusion and employee resource programs and initiatives