Pros
Projects there usually have several phases. First, it is a learning phase, during which you read a lot, while studying previously unfamiliar branch of mathematics. The learning curve is quite steep, but it always feels good to acquire and master new skills on a deep level. Then comes even better phase, when you start sharing your knowledge and ideas with the team, and actually produce meaningful results. After combining and structuring the results, the team produces final report on the project. The project deliverables are typically the big reports, containing the current state of a problem and a bunch of new results and algorithms, that came from the team. Finishing every project feels very rewarding. It's like completing your PhD thesis again, and for that feeling I am truly grateful to the team. I'd love to work with you again, guys!
Cons
Problems come when the management (well, the CTO, actually) tries to get involved into a project. He serves as a perfect illustration of the Dunning–Kruger syndrome, not showing any respect to your and the team's ideas and results, quite often not even understanding them. Sometimes our discussions went south, and he actually shouted at us. Needless to say, that was frustrating, especially because scientifically he not even was incorrect, but didn't make any sense. Though the company is quite small, the culture is far from 'startup'. There is a clear separation between the researchers ('the resource') and the management (CEO & CTO). You'll never get to meet customers, never make any impact on business, never know what happens to your reports and results, never be anything but a kept-in-the-dark researcher. I left immediately, when I was offered similar role in another company performing big data analysis.
