Billions of humans depend on healthy oceans for prosperity and sustenance. Unsustainable exploitation of the oceans exacerbated by climate change are threatening coastal communities worldwide. Accurate and timely monitoring of maritime activity is an essential step to effective governance and to inform future policy. In support of this complex global-scale effort, we built Atlantes a machine learning based system that provides the first ever real-time view of vessel behavior at global scale. Atlantes leverages a series of bespoke transformers to distill a high volume (100M/day) continuous stream of GPS messages emitted by hundreds of thousands of vessels into real-time behavioral classification. The combination of low latency and high performance enables operationally relevant decision-making and successful interventions on the high seas where illegal and exploitative activity is common. Atlantes is already in use by hundreds of organizations worldwide. Here we provide an overview of the machine learning strategy and modeling architecture that enables this system to function efficiently and cost-effectively at global-scale and in real-time.