Any student should be up to their eyeballs reading whatever they can get their hands on related to their chosen field of work. This is even more important when deciding what roles a student wants to aim for after graduation. However, it is becoming clear that even though the term "data scientist" is the hottest buzz word on the market as of this writing, companies are having a lot of growing pains realizing the benefits of data scientist hires - and this is affecting the hires too.
"There is no single job description that encompasses this emerging field. Many people who have traditionally been called 'data analysts' aspire these days to the title of 'data scientist'... and data scientists, in turn, aspire to be machine learning experts."
- Anthony Goldbloom, founder and chief executive of Kaggle
According to an article by the Financial Times published Nov, 2017, only 30% of Kaggle users had any formal college level training in data science or machine learning, while 66% said they are self-taught! Kaggle's Anthony Goldbloom - quoted by Financial Times - listed physics, computer science, classical statistics, bioinformatics, and chemical engineering among the STEM backgrounds that Kaggle users hail from. I for one, am a formally trained biochemist working in computational biology using skills and technologies that I learned through self-study by means of online tutorials, documentation, and massive open online courses. It's nice to be a statistic, I guess.
One of the more poignant observations that Goldbloom has made is that "machine learning is the first new discipline to demonstrate the importance of lifetime learning." I think it's true, especially as millennials take the reigns from baby boomers, that the concept of having a single career over your lifetime just isn't possible anymore. It simply must be accepted that people be prepared to re-skill to adapt as new technologies and opportunities arrive and old ways become obsolete.
This said, lets go back to the term "data scientist" and what it means for hiring managers. Well, actually what it means depends on the hiring manager. There is a severe lack of data and machine learning technical experience at the managerial level right now - and even more so among technical recruiters - and as a result you will see a pretty wide disparity in what the job board listings give as the skill requirements for positions with the title, Data Scientist. For example, I've been told that if I see a job listing requiring experience in Python, R, TensorFlow, PyTorch, NLP, Spark, Hadoop, Hive, Pig, SQL, A/B Testing as well as everything ML (thank you Jonny Brooks-Bartlett), then I should probably run. The advice has been to steer clear of companies that have no idea what they are looking for - no data strategy - as well as those who are looking for unicorns who can fix all their data problems. Actually, a side effect of this is a higher chance of technical recruiters "ghosting" candidates after the first few conversations, but that's beyond the scope of this post - though I might address it in future. That said, I personally imagine it would be easier to build a unicorn team than to find and hire an individual unicorn.
"So what's the big deal?"
I see it this way. Be careful thinking you need to study and develop experience in EVERYTHING that you see in those job listings for what that company needs in terms of skills for the data scientist position. In fact, it might be a good idea to review data science positions at companies who have a reputation for actually knowing what they are doing, i.e. Google, Amazon, and to also directly ask potential employers with not-so-obvious uses for data science what their data strategies are - many companies with loads of data can sense that it can and should be put to work, but have zero idea where to start - and if they already have a senior level data scientist in their employ - companies with zero idea where to start aren't going to get far if they have only a few recently graduated or junior data scientists.
Another piece of advice I was given is that you shouldn't be discouraged from applying to a job if you only meet 60% of the listed requisite skills. Getting that much in a candidate is considered a good start and if that 60% is solid - backed by evidence - the company is more likely to trust that you can fill in the gaps on the job. You can't be expert at everything, but if you have a track record for picking up skills quickly and can show that in production, then you have better chances all around! Being honest about your skills - both to yourself and on your profiles - as well as showing evidence of those skills is key and is the subject of my next post.