Data-focussed roles are nothing new, but the exponential growth in the amount of data collected today has made them crucial hires to most competitive businesses.

For those new to the field, it’s not uncommon to be shocked and horrified by the job ads with an impossible list of requirements. For instance, I’ve seen Data Analyst roles that ask for an MSc or PhD in maths/computer science, proficiency in a laundry list of statistical and machine-learning methods, programming languages, data engineering, cloud engineering and dataviz skills. Add to this, expectations of business acumen, stellar presentation skills and stakeholder management and it’s all a recipe for serious overwhelm!

Google “How to get a job as a Data Scientist” or something similar, and I’m sure you’ll find dozens of articles on the internet. Each one with a prescriptive list of programming languages you should know, skills you should have and courses you should complete. I won’t attempt to reinvent the wheel or reiterate facts you’ve read elsewhere. (That’s not what you came here for anyway!).

However, I will share one of my favourite nuggets from an e-book I recommend to aspiring Data Professionals, especially Data Scientists.

(Just one disclaimer: This is very much geared towards those who spend their time extracting insights from data rather than managing it per se. Maybe one day I’ll understand the scope and subtleties of data engineering but, in the meantime, I won’t pretend to do it justice.)

Right, now that’s out of the way…

If you’re looking for your first role, it’s tempting to dive straight into the code and tools you need to be familiar with. However, at their core, most Data roles fundamentally revolve around a problem that needs solving. Whether it’s mining data for insights that provoke thinking and spark inspiration (analytics), hypotheses that need to be defined and rigorously tested (statistics and machine learning) or analyses or models that are needed to turn information into action (decision science).

Caption: The skills that combine to make data science and how they combine to make different roles. Adapted from Robinson & Nolis – Build your career in data science  

Current data practitioners likely sit somewhere around this triangle depending on the problems that spark joy for them. As a newbie, whether you’re inside the triangle or not, your closest vertex (the pointy bit) is probably the best place to grow from.

For instance, my favourite problems to solve involve asking questions about natural phenomena. I started out with specific domain knowledge in my field of interest – biology. I knew how to design experiments, define problem statements, and formulate good hypotheses. I learnt statistics so I could test these hypotheses and come to conclusions in spite of noisy data (Domain Expert → Statistician). When I was outside my specific area of expertise, I needed to learn how to build dashboards and code to produce great analysis with data collected by others (Domain Expert → Expert Analyst) and prompt more questions or hypotheses.

For those who come from mathematical backgrounds, perhaps solving thorny mathematical problems is the stuff that gets you going. If so, can you simulate real-world phenomena and communicate the probabilities of different outcomes (Mathematician → Decision Science)? Are you familiar with statistical concepts and can you test specific hypotheses with confidence (Mathematician → Machine Learning)?

Perhaps you have a background in software engineering and you’d much rather spend hours solving coding problems. If so, what type of programming do you enjoy? If you’re comfortable with lots of trial and error, can you wrangle some code to run existing algorithms through your data and assess their outputs (Programmer → Machine Learning Engineer)?

If you’re as impatient as I am, or just prefer to have tangible outputs quickly – then can you turn your attention to data exploration and do you know what great analysis looks like (Programmer → Expert Analyst)?

 All this assumes a linear path from one vertex to the next. That’s great if your mind works that way and/or you’re already happy where you are. But if programming has lost its allure for an ex-engineer, there’s not much point waxing lyrical about more things to code in Python when there are no-code AI solutions available.  

Likewise, if you’ve done nothing but write and play music during the last decade, don’t let naysayers tell you can’t write AI algorithms to bring your favourite composers back to life! You absolutely can!

Our individual paths to (data-driven) enlightenment are unique to us. While it’s probably best not to run yourself ragged running laps around the field, a little agility and SMART experimentation, coupled with a few leaps of faith will certainly make for a fruitful and interesting journey!