A Digital Twin (DT) refers to a digital replica of physical assets, processes and systems. DTs integrate artificial intelligence, machine learning and data analytics to create dynamic digital models that are able to learn and update the status of the physical counterpart from multiple sources. A DT, if equipped with appropriate algorithms will represent and predict future condition and performance of their physical counterparts. Current developments related to DTs are still at an early stage with respect to buildings and other infrastructure assets. Most of these developments focus on the architectural and engineering/construction point of view. Less attention has been paid to the operation & maintenance (O&M) phase, where the value potential is immense. A systematic and clear architecture verified with practical use cases for constructing a DT is the foremost step for effective operation and maintenance of assets. This paper presents a system architecture for developing dynamic DTs in building levels for integrating heterogeneous data sources, support intelligent data query, and provide smarter decision-making processes. This will further bridge the gaps between human relationships with buildings/regions via a more intelligent, visual and sustainable channels. This architecture is brought to life through the development of a dynamic DT demonstrator of the West Cambridge site of the University of Cambridge. Specifically, this demonstrator integrates an as-is multi-layered IFC Building Information Model (BIM), building management system data, space management data, real-time Internet of Things (IoT)-based sensor data, asset registry data, and an asset tagging platform. The demonstrator also includes two applications: (1) improving asset maintenance and asset tracking using Augmented Reality (AR); and (2) equipment failure prediction. The long-term goals of this demonstrator are also discussed in this paper.