Introduction
As you consider your first data science job or your next data science position, you will want to ask yourself what is important. For me, I have had a few positions in data science, and these are some of the most critical questions that I think are important to consider when selecting the next job.
Do You Have an MLOps Engineer on Your Team?
Studying data science usually consists of mastering machine learning algorithms, but there is one huge part that is often neglected in academia, which is the operations of these algorithms. The reason can be that there are plenty of different ways to deploy your models, and a lot of the options can include costly, specific platforms that are already integrated into your business. Because of this variability, schools or programs might choose not to include operations on their syllabus, which is understandable.
With that being said, you will most likely want to ask if this work will be your responsibility as the data scientist, or if you will have a dedicated MLOps Engineer (or Machine Learning Engineer, etc.). Of course, there are people who can do both, and prefer to master both parts of the creation and deployment of models process, but it is also okay to be only a data scientist who focuses on algorithms. It is even more important to clear this definition with your future or current manager.
Will You Be Working with a SQL/Data/Business Analyst?
Similar to the above consideration, you will want to ask if you have someone close on your team who is an expert in SQL. Some data science positions require little SQL, while others can require it nearly daily. In your interview, you will want to narrow down how much SQL you can expect to perform, and if you are the only one or not.
Sometimes, there is someone else, like a Data Analyst, Business Analyst, or Data Engineer who is more of an expert, who lives in SQL. However, in some data science positions, you will be required to query your data before the modeling process, as well as after.
Are You Expected to Work on One Project at a Time?
One project at a time might sound like a simple task before entering a professional data scientist role, but it can quickly turn into a full-time project.
You can expect some of the following steps for any one particular project:
- Defining the business problem
- Obtaining the data
- Querying the data
- Feature engineering (not just ranking already existing features, but coming up with new ones that make sense conceptually)
- Model comparison/error/accuracy analysis
- AB/testing model
- Deployment of model
- All of the above can have some type of work with others like a Product Manager, Executive Staff, Software Engineers, Data Engineers, AB Testers, Business Analysts, Data Analysts, etc.
Are You the Only Data Scientist Working on a Project?
Some positions in data science will have a project where only one person works on it, while in other roles, there are a few people working on the same model. People move at their own pace, have more or less productive days over other days, and can enjoy or not enjoy working with others on the same project daily.
It is ultimately up to you to decide what you like, and just as important to know what the expectations are before entering a role.
What is the Usual Timeline for a Project to Finish?
The testing of algorithms/model creation is surprisingly quick. It is the parts before and after that can take up most of the time in developing a model and integrating it into your business.
The timeline can fluctuate for any project, and just as the other considerations above, it is about expectations — how much work is necessary to obtain a useful result.
Summary
Overall, it is important to remember that when you are being interviewed for a data science role (or any role), you should be equally interviewing them back, and these are just some of those questions or considerations you could ask and bring up. Additionally, you could still bring up some of these questions even in your current role.