After graduating from high school and passing my engineering entrance exam in India, a major decision that I needed to make was which engineering specialization I should opt for. It was the era of Dotcom Boom, every other person I went to school with was opting for either software engineering or computer science. So when I announced that I will not specialize in either of these areas, everyone in my family (read my parents) panicked. Opting for an IT related specialization was “fail safe” and in their minds, I was making a big mistake letting go of this fail safe option.
My decision were obviously not based on my ability to foresee the future. I guess I was always a bit maverick enough to not be part of a crowd. Sometimes thinking outside pre-defined frameworks and norms help you get creative and look at things from a totally different point of view. Questioning status quo has always been my habit so I wanted to explore many facets of engineering before I made a decision about the area I wanted to specialize in (I did end up specializing later but not in Computer science or Software engineering).
Then the “Dotcom bubble” burst while I was still pursuing my undergrad in engineering. The key point here is not the gory details of what happened to those millions of “software engineers” after the bubble burst (The real computer and software engineers were still fine). The only reason I recalled that era was because, in my view, something similar is happening again. There is a bubble forming again-this time the bubble is around Data Science.
The exploitation of the Data Science Revolution
I will not waste words on defining the “Data Science revolution”. If you are reading this article, you are probably already aware of how Big Data and AI have impacted every aspect of today’s new economy. With the surge in applications of AI in businesses, suddenly there is now a huge talent gap. Everyone around you is literally shouting in your ears how the world needs more data scientists and advanced analytics professionals (which is absolutely true). However, In my opinion, this sudden surge in demand is the reason things are again going off-track, just like they did with the Dotcom bubble.
Remember the database administrator you know from a job 5 years ago? They now use Data scientist in their LinkedIn profile. Your cousin with a Chemistry PhD who ran SPSS (a statistical programming tool widely used in academia) to analyze algae growth parameters in potatoes in a lab in Idaho ? He is now a “Data Scientist with specialization in bio sciences”.
Every other hardcore IT programmer who has completed an “Introduction to data science with R/Python” online course, claims to be a data scientist. Just learned basics of Sql programming? Congratulations-you can now brand yourself an “Advanced analytics” professional. You can also see “Advanced Analytics” consulting firms of all sizes mushrooming all around you and even pure play IT and business process outsourcing services companies are re branding themselves as “Digital Transformation and Analytics” services providers. It is almost like a feeding frenzy out there.
So where am I going with all this? All this background summarizes into two key takeaways:
(1) Data Science revolution is for real but your organization shouldn’t jump into anything just because everyone else is doing that. Plan and evaluate extensively. Make sure you are ready, you have the right resources in place etc. Don’t just get into the race to be another “also ran”. If Data science would have been only about pure math and programming, you would not have seen some of the leading companies in the world struggling to implement these solutions. A lot can be written about this however this is not the focus of this blog post.
(2) Related to the topic of this post-Be cautious and selective about the Data Science and advanced analytics talent-irrespective of whether you plan to build a team internally or are planning to hire an external consultant. Also, evaluate the different type of advanced analytics roles you need in your organization.
What kind of Data Science talent do companies actually need?
Many of the profiles that are actually being marketed as data science profiles may already exist in your organization’s talent pool. The true Data Scientist profiles though are difficult to find and retain. My perspective is that a best in class data science department will constitute of multiple profiles. Three primary profiles are:
Data Engineers: Few data engineering professionals these days have the professional integrity to use Data Engineer as their job responsibility in their LinkedIn profile. All the data engineers and database programmers have “transformed” into data scientists. For those who still use data engineer title, I salute them for their professional ethics . A true data engineer would have a very strong programming and database design and architecture background and chances are you already have such a pool in your IT organization.
Data Scientists: True data scientists have a very deep math and statistics background, along with strong programming skills (As far as programming goes-may not be as strong as data engineers). Using this deep math and statistics background, they’re creating advanced analytics. On the extreme end of this applied math, they’re creating machine learning models and artificial intelligence. Since the objective most of the time is to help the business, they requires some level of business acumen and communication skills.
Analytics Hybrids/Translators: Some think tanks call them “Hybrids”, others “Translators” . This important new role will very soon become the key to your AI initiative’s success. Try looking up “Analytics translators” online and you will find a plethora of resources emphasizing the need of this role in your analytics departments.
What is the typical profile of an Analytics Hybrid?
Hybrids are neither data architects nor data engineers. They’re not dedicated data scientists as well. Instead, hybrids play a critical role in bridging the technical expertise of data engineers and data scientists with the operational expertise of marketing, supply chain, manufacturing, risk etc. In their role, hybrids help ensure that the deep insights generated through sophisticated analytics translates into impact at scale in an organization
In the planning stage, hybrids draw on their domain knowledge to help business leaders identify and prioritize their business problems to determine which one will create the highest value when solved. Hybrids then tap into their knowledge of AI and analytics tools and algorithms to convey these business goals to the data professionals (scientists and engineers) who will create the models and solutions.
What kind of skills do they have?
- Deep Domain expertise.
- An in depth understanding of the techniques and technologies of data science, along with a fairly detailed understanding of the challenges associated with each (e.g. over fitting, model refresh, challenge of acquiring training data, cost of compute, etc.). They also need to be deeply knowledgeable of the challenges and opportunities unique to each valuable use case, and that means coming from some years of doing hands on advanced analytics.
- A minimum of Intermediate level programming skills in at least one major analytical programming language (R or Python)
- Executive communication skills.
Last, and most important of it all-Where do you find them?
Just like real Data Scientists, real hybrids are elusive and difficult to find. But they are out there and there are approaches you can take to find them and leverage their expertise:
Develop them internally: Tap your internal talent pool. Look for your business line managers with sharp analytics expertise and/or quantitative educational credentials with a high level of interest in Analytics. They may not be initially proficient enough (at a level required for an Analytics hybrid) in advanced analytics but then they have deep domain expertise and they also understand your business. You need to design a training program carefully-you don’t want a curriculum that is too basic or one that assumes that they need to be trained to be Data Scientists.
Hire them (external hires): This is the most challenging option but the good news is that such talent does exist. With all the hype around big data and data science (much of which is deserved), many domain experts and business line managers have developed good expertise in data science skills and are investing their time in learning basic to intermediate advanced analytics skills. An experienced third party recruiter who specializes in analytics recruiting can help you find such talent but be prepared for a long search in this market.
Hire consulting companies: If the plan is to seek help in taking that first step and get help with option 1 (developing internal talent), you may look at the option of hiring an external consultant.
As indicated earlier, the market these days is full of “Advanced analytics consulting” companies. Make sure you do your due diligence before you hire-Data Science is more than databases and algorithm so make sure that your vendor understands your Industry, your business model, the objectives of the analytics exercise etc. Selecting the right vendor is so critical that there can be a whole separate post on this but the gist is-look for a long term partner with a good blend of business and AI consulting skill set. Top management consulting companies have evolved to align with digital economy and are well equipped with a portfolio of talent.
If you are considering launching an AI initiative in your organization or planning to set up an internal advanced analytics team, you are probably doing your research. That means you are aware of how many AI initiatives fail or struggle and what was planned to be implemented in months, may end up taking years (if it does not fail).
A key factor behind these trends is the failure to link advanced core data science methodologies with business objectives and strategies. This gap has led to the rise in the demand for Analytics Hybrids. So far, most of the focus was on developing data scientists to fill the gap but now that some companies have embarked on the journey, they are realizing the missing link. The buzz has already started and soon everyone will be screaming about the lack of true Analytics Hybrids or Translators.