On Linkedin, many Data Science enthusiasts who aspire to be Data Scientists follow me.
One person asked the question:
What do I need to know to get a $150K job as a Data Scientist?
It’s a good question
Because you can find Data Scientist jobs paying as little $30K and also paying up to $150k
So, what’s the difference?
To compound the confusion, recruiters and companies include an assortment of skills in the job description.
Are you really expected to have all these skills?
The secondary question was: Do I need a PhD to get to the 150K salary?
My personal view re PhD .. is not necessarily – but it helps
A quantitative PhD demonstrates that you can solve complex problems. But a PhD is not the only way to demonstrate this. For example, a good Github repository would also demonstrate the same thing (or publishing a paper in a top journal, or a well-known analytics program from a top University like Stanford, Oxford, Cambridge, IIT etc)
So, if it’s not only a PhD – what is it?
Here is my list
It draws from my own experience, that from others, from my teaching at the Data Science for Internet of Things course at the University of Oxford and in helping students to get roles
Note that the views are my own and do not represent the views of any organisation of institution I am associated with
1) The ability to create new Intellectual Property and understand Research papers. In many cases, for more complex parts of AI, you are dealing with some significant unknowns. It then becomes a case of searching Google Scholar or arxiv, trying to find research papers for similar problems and then attempting to duplicate that solution. This is not easy, and it does not neatly fit in the agile / sprint approach also. Hence, it is not a common skillset in industry but is more common in academia. Related to the this, you need to understand the maths behind algorithms because you may need to explore deeper options based on the workings of the algorithm itself if you want to enhance it
2) Working with large and high volume and often real time datasets: If you see the uber tech stack and the problems it is designed to handle, the volume and velocity of data implies that you need to rethink many aspects of the stack. The experience of working with large data volumes and real time datasets will be valuable
3) Full stack experience: In larger implementations, the Data Science role comprises of three roles: The Data Engineer, The Data Scientist and Devops Engineer. Experience of working in this environment is valuable (even if you are personally involved with only one of the components)
4) Large scale deployments: Related to the previous point, as I outlined in the four quadrants of enterprise AI business case, large deployments are likely to involve business considerations like explainability and deployment considerations like CICD(Continuous Improvement Continuous Delivery) etc
5) Selling: This may come as a surprise – but at the higher end of the scale, you would also be expected to sell (probably within a consulting role)
6) Domain knowledge: AI deployments in various domains are likely to be increasingly domain specific. For example, in areas I work with (bioinformatics) – you often have sequence-based data. So, time series algorithms are likely to be used. The actual knowledge of the business problems and the algorithms applied to them will also be increasingly crucial
7) Working with people: You would often need to work with people in context of AI including customers, vendors, management, data engineers, devops, external stakeholders etc.
In my view, the above seven factors would characterise the top end of the payscale for Data Science. These skills often don't go together i.e people who work with maths are less likely to be comfortable working with people
To conclude, and to be a bit controversial:
a) These jobs would need an Engineering or Maths based background i.e. not only an accounting or a management degree with no technical qualifications.
b) They would also NOT be typically offshored
c) They would be permanent roles (not contract roles).
Comments welcome as usual
Image source: shutterstock