A major trend these days is to study Engineering and transition to Data Science at Software Companies. On the other hand, we also have Core Engineers / Scientists (PhD) who have taken up Data Science to complement their research background in helping solve complex engineering problems efficiently by leveraging Machine Learning.
Megha Singh, our next pathbreaker, Data Analytics Science Specialist, develops algorithms for fault diagnosis of Motors and Electrical Machines .
Megha talks to Shyam Krishnamurthy from The Interview Portal about the charm of applying Data Science to Electrical Systems like Motors and Machines and having the ability to not only understand the data but also the underlying diagnostic mechanisms that could help identify a fault, thanks to her Electrical Research background.
For students, don’t transition out of Core Research & Engineering to Data Science. Instead use Data Science to do your engineering job better!
Megha, tell us about your background?
I was a sincere and studious person since childhood. My grandfather was my biggest supporter, he always motivated me to do the best. My father is a Retd. Principal and he used to teach me Maths. My mother is a housewife but has studied M.Sc, so she taught me science subjects. I was also active in school assembly prayers and speeches. I also liked doing yoga and gymnastics which were compulsory activities in our school.
I topped my class in 10th board exams and as I liked science and maths subjects, I opted for these subjects for my high school. In 12th board, my performance was good (topped my batch) but I couldn’t clear any mainstream exams, but secured a good score in the state engineering entrance examination.
What did you do for graduation/post graduation?
I wanted to prepare for the IIT JEE examination for the next year, but due to a good score in the state entrance exam, I got a government seat in my hometown itself. This was a difficult decision and my parents got me admitted to a college in my hometown only. So, I dropped the dream of going to IIT then and there itself. I got admission in Govt. Engineering college through State entrance test in Electronics Instrumentation & Control Engineering. In the final year of my Bachelors, I started preparing for CAT, but soon realized that my interests were in core engineering. This was a self-realization phase for me. I appeared for GATE and qualified it in my first attempt. In spite of being the topper, I couldn’t get placed in college placements. Due to the recession, not a lot of companies were hiring. Later, I thought of preparing for IES but the coaching center was in Delhi. My parents were not ready to let me go to Delhi alone and prepare. So, I thought of continuing my studies and applied for a Master’s degree program in Control Systems in MBM Engineering College, Jodhpur and got selected. My dream of pursuing a Masters in IIT was again trashed. But i did ultimately complete my PhD (Electrical Engineering & Machine Learning) from IIT, Jodhpur.
What made you choose such an offbeat, unconventional and rare career?
In a Master’s program, one has to identify a research problem and give your own solution which is acceptable to the scientific world. My supervisor helped me in identifying one such problem whose solution utilized Neural Networks (these days a part of Machine Learning). I started coding in C and was always immersed in logic and coding. I developed many algorithms in C language as part of my assignments. Those experiences ignited me, I really liked to build logic and code it. I switched to MATLAB and started working on my M.Tech project which was related to compression of ECG signals using Neural Networks. Here, my mentor was Dr. Akhil Ranjan Garg. Another inspiration came from my other faculty in Artificial Intelligence, Dr. Jayashree Vajpai. Both of them always appreciated my efforts and motivated me by giving valuable feedback. Looking at my interests, my guide suggested that I apply for PhD in IITs especially to IIT Jodhpur, where a new kind of interdisciplinary course had started. I applied, cleared the written and then interview. I was selected for the PhD program. No doubt, that was my biggest day and I was looking forward to it not knowing what challenges lay ahead.
This journey has been very unique, there was nothing normal about it. People may get motivated from it by at least knowing what one should not do. I was selected for the Biologically Inspired System Sciences program, where we were taught biology related courses including theoretical and computational biology. My batch mates were from a mixed background; biotech, computer science, electrical, biology. I was totally taken aback by the course content and then the heavily loaded exams at IIT. I initially performed poorly in those exams. That phase of my life was very heartbreaking and I never imagined it. I somehow cleared the courses with good pointers and finally chose to work under one of the best mentors Dr. Ganesh Bagler. The area was related to early diagnosis of schizophrenia using functional MRI data. I started learning and then working towards it. After 1.5 years of hard work, I started to understand this area, with a few marginal outputs. Due to some reasons, my mentor left IIT and I was back to square one as I was caught under rules and regulations of the institute. I restarted my journey, chose a different guide and topic altogether. I started working on fault diagnosis of induction motors.
Initially it was difficult, but my interest towards it made it easier. I was interested in understanding electrical signals from signal processing approaches. I was also interested in applying machine learning in this area. Therefore, I embarked on this new journey with a lot of hiccups which were to come up in the next 4 years. My PhD was on detecting faults and fault types in a three-phase induction motor at an incipient stage (initial stage). We started with a low voltage induction motor, did experiments on it to induce faults at various parts of this motor and recorded the subsequent electrical signals. This particular research on motors is of great importance as far as the real-world is concerned. Around 80% of the industry utilizes the induction motor, and therefore, monitoring the health status of the motor is essential. This takes care of the financial losses, production losses and also helps in planning subsequent standard procedures to be taken after detecting the fault at an incipient stage. My research involved using time-frequency analysis to extract features of the electrical signal in the time and frequency plane and use those features to detect and classify faults with the help machine learning techniques.
After a 2 year journey in IIT, I restarted my PhD and completed my actual PhD in the next 4 years. My interest in what I was doing helped me complete my PhD with my relentless determination in spite of failures of my experiments, extreme peer pressure and the IIT’s innate pressure to extract good performance. My labmates and friends (very few) have been extremely supportive like listening and giving suggestions to at least ease my pain and struggles.
Tell us about your career path
I always knew that I wanted to get into the corporate world. There is a preconceived notion or mind-set that PhDs can only go into academics and become a professor. I was fine being a professor, because I was a faculty for one year and teaching assistant in IIT for five years. I developed teaching skills and had good feedback as well. However, I wanted to pursue a career in the industry. I had no idea as to how to make this happen. So, after being involved in research for some time, after completing my scholarship period, I started looking for internships and jobs. In an IIT, a PhD student gets a scholarship as a teaching assistantship by the MHRD, Govt. of India. To avail this scholarship, we have to perform duties assigned by the faculty for particular courses. This scholarship is given for 5 years. In my case I completed 5 years, after which I couldn’t get the scholarship. Therefore, after some efforts, I got selected in a part time start up venture, which, due to my unprecedented loyalty to my PhD, i could not accept, and hence, did not go. However, I learnt one lesson that my research area is useful in industry and I can always get a chance.
Thereafter, I literally worked hard to finish my PhD and submitted my thesis. After this, I took up a project of my own in IIT which was related to motors but was different and challenging.
My main goal was to go in the industry but I still had no clue. Within six months I applied for more than 150 jobs, post-doctoral fellowships, project positions and what not. I even applied for assistant professor positions which were not a priority for me a few years back. At the same time, I gained knowledge and worked in a project which really helped. This project was related to control of induction motors used in electric 2 wheeler and 3 wheelers. Though this was not directly related to my work, i took up this project to improve my knowledge about induction motors and to be productive during the period which i was waiting for my PhD defence. This project was related to applying direct torque control algorithm to the motor using non-linear control techniques. This area is very relevant to the industry as controllers and drives are the main components of any electrical system.
I also improved my knowledge and acquired skills in both electrical machines and machine learning. With this I got selected for an Assistant professor position in a good private college. I also cleared all rounds of a major research center in India, but due to the lockdown, could not pursue the opportunity further.
Here, I would also like to mention the role of LinkedIn in building my confidence and paving the way to exciting upcoming opportunities. I made several contacts and got some interview opportunities through linkedin.
How did you get your first break?
I applied to many companies and one of them was GE. They shortlisted me for the interview round. In a week, they interviewed me, then another round of interviews, and I got selected in GE PC. They selected me primarily for my data analytics skills in the field of electrical machines. My final round was in Chennai, and I was offered the position. No one would think twice about an offer from a reputed name like General Electric. But I had one more reason to accept the offer, because the work was related to my domain and also would give me flexibility to gradually work on other projects as well. This is what I actually wanted to do.
What were the challenges you faced? How did you address them?
Converting a PhD into a corporate job is very challenging in itself. Many recruiters would not be interested in hiring somebody who is overqualified for a position and then the compensation also plays a crucial role. One of my professors told me that a PhD should get a good compensation because during PhD whatever he/she gets is very minor, just a stipend but he/she survives for a better life in future. So, considering these facts, finding a suitable job is extremely difficult and challenging for a PhD degree holder.
The challenges I faced:
Where to apply? There are many companies and universities where I could apply, but the relevant place of work was most important. Choosing the one which matched my qualifications and knowledge was initially a big challenge because I had to search thousands of websites.
Which role suits me? Having skills is okay, but to understand where you can use those skills other than your research area is a daunting task. Getting to the exact same area was just luck, by chance.
I also faced the dilemma of prioritizing whether I want to go for a Postdoc or for a job immediately.
Where do you work now? Tell us what you do?
I work for GE as a Data Analytics Science Specialist. I have started working on a project which involves fault diagnosis in the machines deployed at different sites of GE. Other projects also revolve around machines like induction motors and drives. The skills required for this job are the same as what I learnt during my Ph.D. The main language I worked on was MATLAB and here as well, my partial work is in MATLAB. However, I have started working on Python as well. As this is my virtual work from home, I keep myself disciplined for the 8 hours during office time. I work and also attend a lot of meetings in between. I love doing what I am doing. I have enjoyed engineering since B.Tech, but I also enjoyed coding and I do both here. My team is cooperative and fun to work with.
Though my major role is in data analytics as the position suggests, I work on analyzing the data coming from motors. Since i have recently joined, I cannot say much. But I would be developing algorithms for fault diagnosis of not only for motors but also other machines. My role is not only related to data science, but also involves working on core engineering problems. Having an electrical engineering background and utilizing data science has a different charm. Understanding the system and then applying data science tools allows me to have a better approach and understanding than a mere Data Scientist. In a practical sense, a pure Data Science professional cannot perform such tasks, they would need field experts with them to make the results and approach feasible. I will take an example using the work I was doing for Schizophrenia. Developing an algorithm to identify if a person schizophrenic based on brain signals is not enough, we have to realize what are the underlying biological mechanisms in the brain that lead to the syndrome. Therefore, making the algorithm biologically plausible is the most important task, without which it wont be useful in the real world. Therefore, having the knowledge of both the system and the tool is very important. Understanding Data Science is not too difficult for non CS people these days due to enormous availability of sources and courses on the internet.
How does your work benefit society?
GE is a big MNC, big name in the energy sector. They provide solutions to ship operators and energy companies. The company has expertise in electric propulsion and automation systems for the oil and gas and power industry. My work may help in stabilizing the performance of machines in these businesses by reducing downtime and thereby reduce the financial costs and provide a safe environment for the people around.
Tell us an example of a specific memorable work you did that is very close to you!
My first work with Neural network for compression of ECG signals was very interesting. We were able to capture the ECG signal through neuron weights and were able to visualize them using Kohonen maps. The compression achieved was also significant. On the other hand, extraction of features from 4-dimensional functional MRI data using Independent component analysis and Principal component analysis and then utilizing machine learning to identify whether the data belongs to a healthy subject or a schizophrenic patient was really very fascinating. Those experiences were my real exposure to the power of data science in the biomedical field. However, my first work which got published in a journal will always remain close to my heart. During this period, I actually understood the extent of work that needs to be done in a PhD to make it impactful. I also understood how beautifully we can deal with signals to extract meaningful information from them. This understanding is not only applicable to electrical machines but also to any field where there is scope for research and innovation.
Your advice to students based on your experience?
I would suggest that you should first gather enough knowledge about what is happening around you, just don’t go on the path that others tell you to. Explore all your options, work on some projects through internships, through projects in your UG and PG courses. This will open up your mind towards different career paths that you would like to follow. This will also develop your skills and build confidence. Make contacts, get information, help others, be humble, stay determined, all these characteristics will keep you in focus and align with what you really want in life. Never leave your instincts, trust your mind. Stay away from negativities that may come from any source, people, actions, your own results. Just keep doing what makes you feel happy and satisfied, you will surely achieve your goal.
I just want to focus on what I am doing currently. I will continue to develop myself as a Data Scientist.
Actually the More exposure you get, the more you groom. Improve yourself and groom yourself to become better.