Nailing your step into data science (and why Kaggle alone won't cut it)

Jordan Newton
Executive Consultant

The data science community is in a weird place right now.

I work the market every day and see the same situation across the whole of the USA.

On one hand, there are hundreds of fresh graduates and aspiring candidates who ooze potential struggling to get anywhere with the very few entry-level positions available.

Meanwhile on the other, hiring managers are desperate for experienced data scientists and we are seeing one of the biggest talent gaps Darwin has ever had the privilege of bridging.

Whilst I spend most of my time connecting experienced DS professionals with the top companies that need them, my heart goes out to all the great candidates working day-in and day-out to secure their first data science role.

I wanted to share some advice on how to nail that first step, both from my perspective as well as from those in my network who have pulled it off.

Here we go:

Start with Kaggle

A platform for analytics competitions and predictive modelling, Kaggle allows existing and aspiring data scientists to demonstrate and hone their skills by solving tough problems across many industries.

Solutions must be highly accurate and to win competitions, entrants have to have a deep understanding of which solution to apply where (i.e. which algorithm is best for the task at hand). 

However, some have argued that for data scientists looking to enter the industry, Kaggle competitions aren’t the be-all and end-all.


Because a lot of the hard work has already been completed. You'll rarely find a real-world data science project that tells you what the exact question is, as well as giving you the precise data you need, already collected and cleaned. This is exactly what Kaggle does, and so focussing entirely on the platform to give you an edge over your competitors might cause you to miss some major elements of the data science project life-cycle.

So what's next after Kaggle?


Actually do data science 

Instead of relying exclusively on platforms like Kaggle, to become a data scientist you need to do data science. Identify a question in your own life or work that you believe you can answer with existing data (or data you can find) and create a strategy that enables you to answer it.

This is a lot harder than it sounds at first, but once you get to the point of solving the problem, you’ll have a broader understanding of the preparation required throughout a DS project, both in terms of identifying the right questions to ask, as well as going about collecting and cleaning the data.

Sounds like a good place to start, right?

A data scientist looking to stand out from the competition also needs a strong business acumen and an understanding of how their work will be able to impact the bottom line.

It will do you no good having an impeccable technical understanding if you’re unable to tell your interviewer how you plan to implement that understanding to make business change.

The second most common keywords for Data Science job postings is business:

It's very important to companies that you understand the business side as well.

Most aspiring data scientists I know spend their evenings with online courses and reading books, and while this is vital; maybe add a book about marketing or business strategy to your list, or do a MOOC on operations or finance, anything to round out your understanding. You’ll thank me later.

The next one is the easy part.


Demonstrate your passion

Nailing your step into data science requires you to trust your ability to implement, and not just your knowledge and understanding.

Remember, after you’ve entered a few competitions on Kaggle, solve a problem of your own and demonstrate the full life cycle of a project. If it’s a project that could have business implications, even better.

Once you have solved the problem you identified, add it to your portfolio; but don’t stop there, tell someone about your project, your mum, your dog, even yourself in the mirror.

Get used to explaining succinctly how you identified and answered a question with data, and by the time you come to telling an interviewer about it, the job will already be in the bag.