Data labelling capabilities/facilities are specialized systems or environments integral to the AI and ML model development lifecycle. These capabilities/facilities are designed to generate high-quality labeled datasets, which are crucial for training, validating, and testing machine learning algorithms. They span diverse workflows, integrating human expertise and automated technologies to ensure efficiency, accuracy, and scalability. Supporting data across various modalities such as text, images, videos, audio, and sensor data, these facilities apply metadata or other annotations to curate datasets necessary for supervised and semi-supervised learning. Depending on the security classification of the data, labeling operations can occur in secure physical labs or cloud-based environments, employing advanced encryption, anonymization, and access control mechanisms to protect sensitive information. The architecture of data labelling facilities integrates automation, human labor, and quality control to maintain accuracy and consistency. Automation through pre-labeling algorithms and active learning accelerates the process, while human involvement remains critical for tasks requiring contextual understanding or subjective judgment. In-house teams often ensure higher security and compliance for sensitive projects. These capabilities/facilities also seamlessly connect with broader AI/ML pipelines, leveraging feedback from training models to refine labeling priorities and enhance data quality iteratively. By combining robust infrastructure, advanced technologies, and skilled human intervention, data labeling capabilites/facilities enable organizations to develop high-quality datasets essential for building and refining cutting-edge AI and ML solutions.