I'm one of two members in a newly developed R&D arm of our company, developing machine-learning based products. R&D by it's nature has slightly unclear goals, timeframes etc., as you're not sure what will be useful, or when. On the other hand, production is very deadline driven, with clear outcomes. I'm looking for suggestions as to how we integrate both sides of this divide - R&D will have a better understanding of the problem space, while production will get a chance to consider different approaches to problems etc. Any good resources/ideas would be welcome!