One of the things they didn’t tell you when you started doing Data Science (DS) courses and MOOCs, is that a lot of your time (a looottt) will be spend in meetings.
These meetings are important. Very important. There you can understand the business, the goals of the area, their KPIs, and what are the requirements for the work they want you to do.
Sometimes listening it’s very hard. You may hear some things that you don’t want or have all your beliefs shaken.
For this reason, learning to listen is a skill that is acquired over time. Commonly in work meetings, there will be many different points of view, and instead of trying to impose your point of view always, it is better to try to reach agreements and more general solutions that solve the problems.
Careful here, that doesn’t mean that if you are right, and you have the means to prove it, you should just stay there and agree to whatever. The concept of idea-meritocracy is important here. Everyone has a point of view, one better than the other, being able to discern and find the best solution to a problem is possible. Here you can see a great video that explain this in a more graphical way:
Be able to listen and understand it’s crucial if you want to add value and improve a process in DS. So here I mention some of the things I learned (the hard way) on how to listen and behave in these meetings:
Yes, you may think you know a lot, or that some of the models people created before, because the don’t use “Deep Learning”, are not enough. But that’s no the way.
Listen what they have to say, understand the mental processes they went trough to create the models and solutions. And don’t underestimate, or say “yep, the things you guys did are old and weird, wait for mine”.
Don’t just be there staring at the presentation, or at people faces, your phone or anything else. Take these 30–45 mins to have a productive meeting and focus. Take notes if you need them, but pay attention, they deserve it.
That means, ask questions, be interested in the things they say and do. Build, don’t destroy.
A good data scientist needs to transform problems and ideas into well-posed questions, with that I mean a question that has a solution through the data science process.
If one cannot find a viable path to solve this questions, that ultimately will solve the problem, then there’s two options, go back and keep asking questions, or the problem cannot be solved with data science.
Some example questions you can ask that will help you clarify the air in some meetings:
When was this model created and what data did you use to build it?
What are the restrictions of the models, and the assumptions you made to create it?
Which languages or frameworks did you use to build the model?
Do you have a complete documentation of the process you did to create the model?
How is the model performing? And what are your needs for the data science team?
How many people are involved in the project?
Why did you take this path? Are there more you considered?
Where does this variable come from?
Can you share the technical details of the project with us?
What is the scope of the project?
What is the timeline you propose for the project?
How flexible are you with the objectives of the project?
Why do you think this is a Data Science project?
What is the priority of the project?
This is your time to shine and help!
Thanks for reading. If you have any suggestions or other recommendations please share them :)