Superconducting devices are one of the most promising platforms for building quantum computers, but demonstrating fault-tolerance on any quantum computing implementation remains a challenge. In recent years new hybrid quantum/classical algorithms targeting near-term devices have been proposed, focusing on short-depth parameterized quantum circuits and using quantum computation as a subroutine embedded in a larger classical optimization loop, without the immediate need for fault-tolerance. Rigetti Computing has built a flexible computing platform targeting precisely such hybrid applications, relying on custom entangling gates based on parametrically-activated interactions. In this talk I will explain the physics behind these two-qubit gates, how it enables the implementation of two distinct classes of entangling operations, and describe many features that make this architecture attractive from a scalability perspective. Finally, I will present how this gate architecture was used to demonstrate a hybrid algorithm for an unsupervised machine learning task known as clustering on a 19-qubit processor.