As we continue to operate our aging pipeline infrastructure, problems caused by their leaks and the resulting spills also continue to grow. Today a typical leak is not found by the systems deployed to detect them. Instead they are found by individuals - usually third parties - that visually observe the spilled product. The failure in detection leads to large spills with large associated costs, both in terms of third-party damages and in terms of reputation. A new external leak detection system was recently developed utilizing industrial internet of things (IIoT) and machine learning technologies. The system can be deployed without contact with the pipeline, and hence easily installed on an existing underground pipeline. Small (1%) and large leaks are detected utilizing the acoustic emissions from the integrity breach. Field connectivity for the devices are provided through a proprietary long-range low-power network. This allows for data analytics to be performed at the system edge and on a central server to drastically minimize false positives. The leak detection system was recently deployed on a pipeline and successfully tested with induced leaks. With real-time monitoring and +/- 5 feet location accuracy, the system allows its operator to find out about leaks and implement an expeditious emergency response plan, preventing large spills.