Hi LinkedIn,
I have recently received many questions about how I transitioned into the Data Science field. I am writing this post to detail my journey into the world of data science as a graduate in Chemical Engineering and Applied Math.
Initially in my journey, I was intrigued by the prospect of a career in data science. It seemed like the perfect career for someone with a quantitative background looking to transition into the tech industry. So, I set out on my learning journey and tried to learn more about the field. From my initial impression, the field seemed to focus on applied statistics. I believed that the main purpose of the field was to use statistical/mathematical problem-solving, and the software tools (Python/R/SQL) were merely tools to achieve that. This led me to go back and review many statistical concepts like distributions, p-values, correlation, and other things I had learned in the past. The majority of the data science education I found really focused on this aspect of data science.
Then, I started to learn about machine learning algorithms. Starting with the basics like linear regression, logistic regression, and k-means, etc., I found it very interesting to learn how supervised, unsupervised, and reinforcement training worked. I even delved deeply into deep learning to see how some popular neural network architectures worked.
During this whole time, I was also developing the tech skills used in data science, like learning about databases, cloud services, data engineering, programming, etc. I then had a realization that 90% of the actual work in data science is the software developer/engineering portion. This is the point where I switched my attention to the software aspects like programming paradigms, containers, data structures and algorithms, OOP, cloud, CI/CD, Infrastructure as Code, system design, etc.
It seems to be part of a much bigger trend where data scientists are going to be expected to be able to write production-ready code. This is something that the education has not caught up with and might come as a shock to aspiring data scientists.
It's not that the math and models are not important, but that the software skills are paramount for functioning day to day before you can even think about the math and models.