Webinar on Data Science jobs across geographies - Text Synopsis and Video

Recently we hosted a webinar to discuss the current plane of the data science industry and where the trends show it will land because as data scientists we can’t help but validate trends! :) We looked at the industry from two perspectives, one that of the talent: the data science professionals who yearn for projects that will drive them, and one that of a CEO/Manager/Company's perspective who have big-picture ideas that need to be executed with thoughtfulness. Our general theme was on delivering value through barriers like geography, unstructured communication, and general inability to work together in tandem. We also spoke about the gig economy in data science as opposed to full-time roles and what each has to offer. 

Some questions we discussed were “application of Deep Learning in data preprocessing" asked by Reynold Dass (https://www.dhirubhai.net/in/reynolddoss/), Ajay (https://www.dhirubhai.net/in/ajayohri/) answered this with his years of practical experience by actually mentioning that only specific problems are ones on whom any kind of neural nets can be used and about how much data is required in case one wants to use a deep neural net. A couple of questions we heard around the career path (more like an ocean) of data science.

One such question was from an experienced data science practitioner, Nishant Gautam (https://www.dhirubhai.net/in/nishant-gautam-70382281/), who after working for a while in the industry noticed - “With every week that we move forward today, I see a lot of new methods, techniques, and architectures in machine learning space. These are mostly coming from the academic world.

However, the problems that I have solved over the past 2-3 years. Around 70-80 % of them get solved by fairly straightforward techniques, such as SVM, Multinomial Naive Bayes, Logistic Regression, etc. (Given we have a good understanding of the data and problem)”. 

A seemingly different question around ‘gig or stable job’ turned out to be extremely related to this though. The answers as chipped in by Ajay and Byron (https://www.dhirubhai.net/in/byronfuller/) both were very similar. As it turns out, most common problems do have a simplistic existing solution, and engineers are required to build a system easy to consume and slightly fine-tuned to the exact case on top of such a solution. This is what you can expect in a gig economy. It has its own advantages, of course, change of people, self-decided working hours, etc. But to apply data science in novel ways you tend to be working as a full-time employee - one has to understand the problems that are not easy to define, this generally takes months at any given place and is often more satisfying than rote data manipulation you’d see in smaller gigs. An interesting thought rose from the former question that if most of the work is repeatable, soon that bit will be automated and humans will have to solve different problems or the far more nitty-gritty problems related to the current ones. 

We also spoke about how to access meaningful full-time remote jobs that solve novel problems in Data Science. What are your views on that? Share with us, and stay tuned for the next webinar.