How I discovered my passion for Data?
Transitioning from Mechanical Engineering to Data Science
Growing up, I always saw people working at 9 to 5 jobs in tech as uncool. I never thought I would enjoy tech. Yet, like most confused souls, I chose Engineering, Mechanical Engineering to be precise.
However, it didn’t turn out as bad as I expected it to be. Although I wasn’t much into the core subjects, I thoroughly enjoyed my Mathematics classes. I knew I wanted to pursue something around Math but had no clue what. Eventually, the master’s in Applied Mathematics or Statistics degree fascinated me. But, I was delusional due to limited career options moving forward.
One day, my father introduced me to data science and suggested it would be a great career option, considering you like Math. Excited, I did my research. My excitement went from a hundred to zero as I discovered Data Science requires excellent programming skills. I despised coding, provided I had never actually learned it properly. I wasn’t sure if it was my cup of tea. I moved on.
Moving forward, being a part of the database committee for our college introduced me to data. For my first hands-on experience, I collected volunteer data to figure out how many hours they had worked in the event, whether they are eligible for a certificate or not, and a lot more. I bunked my lectures and gave it my all to avoid mistakes. But, guess what? I messed up. People came barging at me, and it was an experience of life. I screwed up again. But, this time, I heard a voice inside me saying, “But, it was fun.”
Following the voice in my head, I decided to take up online courses to feed my curiosity. I started with the ‘Data Science Math Skills’ course and next, learned advanced excel. And no! I didn’t jump to coding yet, but it was a solid first step. Fortunately, in my final year, I got to choose data analytics as one of my elective subjects. And then, I was left with no option but to learn how to code. Few months into studying the curriculum, I was fascinated and realized how much I love being around data.
Our professor assigned us a Turbine Failure dataset and gave freedom to explore and come up with as many insights as possible. I wanted to give it my best shot. So, I worked hard to study the dataset and skills to implement different techniques. My efforts paid off since the external examiner acknowledged my performance and recommended me for a part-time internship at a company. The company hired me as a project intern for the Development of an ETL Pipeline in Python. I learned about data cleaning and transformation to convert highly unstructured data into a clean and organized dataset. My skill set and knowledge of programming grew exponentially. Although occasionally coding was uncomfortable, I felt confident.
I relished the thrill of getting a code right after 100 failed attempts.
Additionally, I also interviewed for a start-up company that intended to take up a project in the failure detection domain. Finding Data Science solutions for Manufacturing problems was the essence of my work at this company. It was perfect as I got to blend my passion for Data Science with the Mechanical Engineering knowledge gained from my undergraduate studies. Working at this start-up company was challenging and enriching at the same time.
I was assigned a research-oriented project on a 13 GB dataset collected from the sensors mounted on the motors. The aim was to build an ML model to predict the type of fault in the machine. I had to figure out everything from project workflow to the actual execution of the project. To begin with, I started with reading a lot of research papers and literature reviews to get a sense of direction and learned the theoretical part of vibration analysis for fault prediction and the techniques like Fast Fourier transform and signal processing in Python. When it came to actual implementation, it was a real challenge, and there were a lot of times when I felt stuck. However, my mentor always pushed me to learn and implement things out of my comfort zone. Finally, I completed the feature extraction and built a Support Vector Machine algorithm to predict the faults. The model accuracy came out to be 98.67%. The project got selected for further development, where I collaborated with an IoT engineer to deploy the ML model and take it forward. I also worked on two customer projects, several case studies of IoT collected data, and short research on graph databases and their application in the manufacturing domain.
Starting as a novice, I emerged as a completely new person. Through the journey, I fell in love with data. For the first time in my life, I knew I was on the right path. The aversion for tech and 9 to 5 jobs faded away, and I found my passion!
Thanks for Reading!
If you liked my work and wish to support me..