Small particles in the atmosphere can impact cloud properties, such as how much sun is reflected by clouds, if a cloud produces rain, and how long clouds last. The particle-cloud interactions that determine these different properties are complex and difficult to understand, and we need more information about the amount and location of such particles to understand them better. Here we use two remote sensing data sets and a new machine learning algorithm to understand how smoke particles are interacting with a large stratocumulus cloud deck over the Southeast Atlantic. The development of this method will enable future spaceborne assessments of particle-cloud interactions, which means that such observations can be made across the globe and over long time periods.