Example: A Machine Learning Team
There is a team lead, who is also Chief Scientist of the company. She built highly effective vision recognition models for a “unicorn” company, and then was recruited to work here. She is very intellectual, and is somewhat impatient with others.
There is a co-lead who is a natural servant leader, and also is naturally Socratic. He gets to know each team member as an individual, and stays on top of what everyone is doing every day. He is also intellectual, inquisitive, and supportive. The rest of the team interact mainly with him rather than the primary lead, who is fairly hands off, but she is always present for discussions about model changes.
The team makes its latest working model available to a set of product application teams, via a git repo’s main branch.
The team maintains models under consideration in feature branches. There are also some other repos that contain models of other architectures that are being tried to see if they perform better.
The ML team held a workshop with the dev teams early on, to explain the model, and work through how to integrate it into the product.
The architect of the application product collaborates frequently with the ML team, usually the co-lead, to work out integration issues, including how to provide feedback on the tests that the application teams run. The architect has a working understanding of ML neural network architectures, which is necessary to enable collaboration.
From time to time the product lead arranges a demo and deep dive of the application, and invites the ML team. The goal is to enable everyone to see the whole product in use, as the product evolves.