How do you organize a data science team process?

Working in R&D can sometimes be a pain in the a*, especially when designing a process that would fit the whole team. I've heard about and have worked on projects where there were no daily meetings, no discussion about technology across the team, and it was kind of free-form till the due date. Enforcing rules on data scientists often backfires with their productivity too.

How do you tackle such a problem in strictly research & development areas?