How exciting, after all that study, spending umpteen hours watching MOOC videos, and praying at the Andrew Ng statue you have in your room, you’ve finally landed your first illustrious role as a sexy Data Scientist, congratulations!

Hmmm… now what?

To get you started on your journey, I’ve compiled some wisdom that I often share with junior members of my teams, and Data Scientists I mentor.

Why?

From day one, the most important thing to do is to LEARN WHAT THE BUSINESS PROBLEMS ARE!

Remember, as a Data Scientist you’re there to help solve problems, and specifically, to turn data into insights, so focus on the business problem first and foremost, then figure out the easiest and fastest way to solve it using your technical prowess.

To do this effectively, you’ll need to befriend the domain experts, business SME’s and front-line staff, who have intimate knowledge of the organisations strategic goals, needs and struggles. Learn to speak their language, understand their pain, and then, figure out how you can help make their lives easier ie maybe automation, possibly new ways to visualise their data, maybe even do some predictive analytics for them… this is when the fun begins.

Always start by asking WHY, and questioning why things have been done a certain way for so long, and always LISTEN to the answers and don’t dismiss them, but offer new views and approaches.

Where’s the Data?

You’re of no use as a Data Scientist if you can’t first find some data! believe it or not, it won’t always be handed to you, so get ready to start searching, and sometimes, begging to get access to it.

It’s important to build strong working relationships with the data custodians, as they have intimate and tacit knowledge of the data, business rules that may govern it, and how to interpret it. Be warned that you won’t always have associated documentation, schemas etc, so they’re knowledge and expertise can make your job a whole lot easier.

Look at me, I Model & I Train!

Ok, next it’s time to do some REAL work. This is where all your coding skills, and hours learning about how Deep Learning really/supposedly/surely works, come into play…

You’ve found a problem, you think you can solve it via a RNN, so you start coding away, copy-and-paste from Google, consult StackOverflow, …

Whoa! Hold on.

First, ask yourself, do you really understand the business problem? If so, does it really need to be solved by a RNN? – Really? So how much data do you have/what will you run it on? What are the business ramifications for a RNN vs another model?

It’s always best to start with the simplest model first, and only then increase complexity as required. They’re much easier to understand, explain and validate.

But, can you really Code?

Sure, you may be able to string together hundreds of lines of Python that train a CNN and spit out a result with decent accuracy, but how easy will it be to reuse parts of your code for the next project, will a colleague be able to debug it when you’re on leave, could the same model be written in half the code, how much testing have you done…?

An important aspect of being a great Data Scientist, especially when working on production-level code, is to employ software engineering best practice, and to understand how different coding paradigms work, as I’ve previously discussed in a post.

NetWORKing

Yes, being able to solve business problems by using the latest and greatest algorithms is imperative to your career, but what will really help you progress and develop, and land that next exciting and challenging role, is networking.

Oh no, I hear you say, I don’t want to stand around engaging in small talk, when I can be at my desk hacking away at my code!

Yes, I understand, it can feel forced and tiring.

However, it’s invaluable to growing your career!

The trick is to attend gatherings with like-minded people. A great way to do this is to start attending local Meetups, and even consider starting your own!

Don’t forget to attend domain specific networking events too, related to your industry. You may initially feel a bit out of place, as you can’t easily geek out with other Data Scientists, but they’re imperative for helping you gain a deeper understanding of your industry and it’s inherent problems/opportunities, which often lead to new insights and applications of your technical skills.

Who’s My Yoda?

To help boost your career, I highly recommend seeking out at least one mentor.

A mentor can be the key to finding that next awesome role. Not only can they guide your career, but they can also leverage their network to connect you to potential employers, and other people in a similar stage in their career.

Some final Tips

  • Use less PowerPoint slides and More demos – use visualisation as much as possible with stakeholders
  • Drop the technical jargon, and learn to speak the business lingo
  • Show your passion for Data Science, and excite/educate others
  • Never stop learning, and try to understand the fundamentals as much as possible
  • Never stop asking WHY!

Now, go forth and prosper…