generalist or specialist? choose between startup & enterprise
The main difference I noticed between being data scientist at a startup and big corporation reduces to the same topic of "generalist versus specialist" or "T-shaped versus V-shaped".
Let me elaborate on this.
big corp
At the big corporation, like Volvo Cars, there are so many people in the team that you don't need to know everything.
You will have a team of Data Engineers to build data pipelines, senior Data Scientists with PhD to challenge and mentor you on the accuracy and complexity of a model, Machine Learning Engineers to deploy the model that you have just trained in Jupyter notebook.
In practice, your scope at big corp most likely be building a dashboard, data preparation, some other small piece of the data pipeline or train a ML model (if you are super lucky).
All of it within the tech stack that was chosen long before you. Using the processes and methods chosen by someone and passed to you.
Given the size of the team, it is just natural that either you are hired as a specialist to do very specific thing or you will become one.
You are part of the team, so expect a lot of meetings and collaborations. Yet it would be hard to have big picture and almost impossible to influence the strategy. Everything will be moving slow, no expectations to finish it "yesterday". A project could last more than 6 months and that would be fine, because we don't have luxury of fail and it should be perfect.
startup
Completely different situations if you are in a startup.
First, you will be interacting with everyone and that's amazing.
Second, you won't have a luxury of time.
The startup has to test as much hypotheses per period of time as possible. This implies that almost everything you deliver as a data scientist will have to be delivered fast (within weeks), even if it means not perfect. Everything that improves the status quo is good and should be released. If there were nothing before and you are doing zero-to-one projects, then literally everything is good.
Most likely you won't have infrastructure setup, nor you will have many data scientists nearby. You are encouraged to suggest and influence what tech stack to use, what methods and processes to setup, what infrastructure should look like.
More importantly, most likely you will have to do data engineering, analytics engineering, analytics or dashboards, ML training and deployment yourself. You will have to be generalist.
You will have to learn just enough to deliver a project, so you won't become a specialist (yet).
