Agile project management is an iterative approach to planning and managing a project. With other project management approaches in software development, such as the waterfall model, you typically identify all the requirements upfront and plan a project meticulously from the beginning to the end. In contrast, in agile project management you do not set a project plan in stone.
Instead, you break the project down into small parts and goals that are achieved in small sprints lasting anywhere from a day to a few weeks. That way a project is delivered incrementally and can be tested and adjusted frequently so that new requirements can be identified.
Agile project management has turned out to be an extremely useful approach in implementing ML projects. The main advantage of agile project management is that it is very flexible and so it can accommodate unforeseen changes, risks or issues that might come up during a project. This is the right setup for ML projects, in which – unlike projects in other areas – uncertainties are inherent. There are a lot of things you do not know beforehand. How good is the data quality? Can the process be predicted with data? Are models sophisticated enough to fulfil their goals? And so forth. With agile project management, answers to these unforeseen questions and other issues can be more easily accommodated.