Data science projects vary greatly in scope and structure. And, while there isn’t a single way of conducting such projects, they often share similar workflows. For example, every project must go through an integration phase, whether it’s carried out by the data scientist herself or by handing the integration off to an engineer. Project workflows are full of decisions regarding complex interfaces (programmatic and organizational) and resource assignment (human and computational).

Using Gong’s research projects as a basis for our discussion we’ll sketch a typical project workflow, describe the various stages, and suggest additional ones if necessary. We’ll identify the potential bottlenecks and inefficiencies and propose ways to overcome them. At the end of the day, how do we ensure that we deliver a relevant product as smoothly as possible and in the required time frame?