Three Months into Data Science Career

I just joined Visa on the Technology team as a data scientist since September 2019. It’s been three months since I got back into the real world and a real job (haha). But the learning doesn’t stop quite there. Rather, I discovered things that differ quite a bit from the academic environment that surrounded me while on campus. I am sharing some thoughts that I have since I started my job. For those who are hoping to enter the data science world, I hope that you might find it comforting that even on the job, you never fully figure out the best solution or the best accuracy rate because the real world is a FUZZY mess! But that’s also what makes it fascinating :)

• Google is your best friend, next is your colleagues

Knowing how to search and what keywords to search for your problem is an art. Unlike homework, there is no one right answer, sometimes answers pop out indirectly. Methodically peeling away the layers of the problem and look for answers methodically may help you solve something you originally had no ideas altogether. Colleagues can point you to company-specific resources or weird confgiurations that only your company would have.

• Familiarize yourself with engineering best practice

When I was at Georgia Tech, we leisurely used git but pretty much pushed everything to the master branch. The real world is much messier than that!! There are multiple branches, you may push and pull and checkout certain files only. LEARN VERSION CONTROLS PROPERLY.

Another thing is learning basic Unix commands, bash shell scripts, and vim editor to read and edit things on the fly will go a long way since a lot of files live on servers that you may never be able to download on your local machines. Be comfortable to work with files remotely.

Lastly, proper documentation is everything. It is easier for you to pass on to the next person and also for the next person to pick up much faster. Next time if you find yourself some idle time, make some diagrams, write out some process workflows on the wiki page, and you and your successor will thank you in the future.

• Scoping and be flexible

The real world is fuzzy and uncertain. It is not just an Jupyter notebook asking you to build a Support Vector Machine algorithm. Sometimes, it takes business users more time to figure out the exact use case or how to use the data. Please be patient and learn the process to brainstorm and pivot along the way.

• Take two and give one!

Every one only knows parts of a puzzle. We can help each other out. Don’t let your imposter syndrome stop you from contributing. Even I was new on the team, I am already leading a small session to teach my teammates how to analyze complex transaction data. You are stronger than you think!

This is all the musings I have for today. See you next time.