Since their introduction, synchrophasor measurement networks have been steadily growing and are expected to continue growing as distribution Phasor Measurement Units (PMU) become mainstream. Moreover, recent advances in data-driven methods, in contrast to traditional synchrophasor applications, introduce more demanding data requirements. Together, these factors create the need to research data engineering methods for handling synchrophasor data at a large scale. This research investigates the use of a stream processing framework to build synchrophasor applications and proposes a novel reconfigurable synchrophasor data pipeline for near-real-time applications. The processing pipeline is part of a larger system which comprises data collection at the substation level, data ingestion, processing and a centralized metadata registry. In addition, to overcome some of the technical, cybersecurity and economic challenges of traditional PMU networks, the system is developed using open-source, off-the-shelf technologies and is deployed to a cloud provider. Lastly, the system is tested using a real-time simulation, the data is collected by multiple agents and forwarded to the cloud using different data rates to assess its performance in terms of the latency.