Recent years have witnessed a massive growth of mobile and internet-connected devices being embedded in our cities, factories, homes, vehicles, and even bodies, and generating unprecedented volume of data. The synergy of machine learning capability with the Internet-of-Everything can leverage knowledge discovery and promote a new horizon of applications that can smartly sense the surrounding and dynamically interact with other systems and users. However, the realization of emerging AI techniques requires large-scale data, is computation-intensive, and is not suitable for training on devices with limited power, processing, and communication capabilities. This talk discusses various issues, application opportunities and challenges related to the Internet-of-Everything. It will also discuss different architectural models and enabling technologies, hybrid techniques, and future directions for edge intelligence and collaborative learning environments that can overcome the computational bottlenecks, reduce latency of decision making, and enhance scalability and privacy.