Beyond The Semesters: E04
Welcome back to Episode 4 of Beyond The Semesters, an initiative by Coding Club, IITG in collaboration with Student Alumni Interaction Linkage(SAIL), IITG where we interview and try to get a peek into the life and work of IIT Guwahati graduates currently working in different domains of computer science.
This time we have with us Aditya Guglani from the class of 2014. He is a Mechanical Engineering graduate and currently works as a Machine Learning Engineer at Facebook. He has also worked as a Data Scientist at Uber from 2017 to 2020. He has obtained his Masters in Machine Learning from Carnegie Mellon University (CMU). He also interned at PlayStation during his MS at CMU.
He shared with us many insights into the industry and gave valuable pieces of advice for all the students who wish to excel in the software industry.
The following is what beholds beyond the semesters !!
- What does a typical day in the life of an ML Engineer at Facebook look like?
The daily routine varies according to the stage of the project which we are working on. In a product-based company like Facebook, the problem starts with the product. We work with Product Managers to understand the problem they have identified. Our main goal is to convert this problem to an optimization problem that can be solved using ML techniques. In the initial stages of any project, we have a lot of meetings to design the timeline and roadmap of the project. In the second stage, we work on developing ML pipelines so that we can create new features and do some experimentation. The third stage is the fun part wherein modeling, predictions, and research starts.
2. You have worked as a Data Scientist at Uber from 2017 to 2020. What made you switch to Facebook after 3 years at Uber?
There were multiple factors. Even though I was a Data Scientist at Uber, my work was similar to that of an ML Engineer. I was looking for a new challenge in some hard-core ML domain. The scale and complexity of ML problems solved at Fb are way higher compared to that at Uber. For example, Fb has a really complex recommendation and ads ranking system. The scale and magnitude of these problems were exciting for me. In recent years, the leadership at Uber has started to switch its focus from Research and Development to profitability. This meant that the investments in AI and Data Science were reducing. This pushed me to look for newer opportunities where I could do exciting work.
3. You interned as a Game Specialist Intern at PlayStation while pursuing Masters at CMU. What did you work and how did you find that experience? Did you also work on game development?
I was involved with the team working on Analytics. They have a separate team that works on game development. We had to analyze and understand the user behavior while using PlayStation. For example, how much time users spent on a particular game, their game behavior, their journey of navigation on the platform, etc. Using insights derived from analytics, I had to propose ways to make the ShareFactory App better within the PlayStation ecosystem. Apart from work, the team also had a lot of fun activities planned. We used to play PlayStation during our free time and the company also organized art and sports events.
4. What motivated you to pursue higher studies in a field completely different (i.e. Data Science) from your major in B.Tech? (follow up: why pursue higher education instead of a direct job?)
It was primarily because of my one-year stint at a data science consulting firm right after my undergrad program. My B.Tech was in Mechanical Engineering and I started working in analytics consulting after graduating. It was during this time that I was introduced to the field of Data Science/ML. I realized that ML is a leveraged domain where it can create an outsized business impact, and as the quantity of data and computing grows, the leverage will keep on getting bigger. Also, I really enjoyed using my quantitative skills to create this outsized impact. Even though my bachelor was in a quantitative domain, I realized that I need to pick up some advanced core ML skills and that’s how I decided to pursue a Masters degree.
5. Can you guide us through your application process while applying for Masters?
I appeared for GRE as most universities ask for your GRE score. I mainly focused on the Verbal Reasoning section as the Quantitative Reasoning section is relatively easy. Preparing with my batchmates helped because we would regularly check each other’s vocabulary (A good vocabulary is required for the verbal reasoning section). I prepared for around 2 months and gave two mock tests. The mock tests helped me to assess my weaknesses and I worked on them in the days leading up to the exam. I scored 320+ out of 340 which is good enough to get you into the top universities. For the second part of the application process, you need 3–4 strong Letters of Recommendation from your professors in college and/or your managers at work. I reached out to two professors from IITG; one being the professor whom I had worked with for my final year BTP project (based on fluid dynamics) and the other being the instructor of the course where my attendance was unusually high (sic). For a strong letter of recommendation, the professor should attest to your ability to identify and solve difficult problems even if that is in a field different from the one in which you wish to pursue your Masters. The third and most important part of your application process is the Statement of Purpose (SOP) where you should have a story to tell about why you wish to pursue Masters in that particular field. Having a good CPI also helps but it is not the only important thing in your application. One thing that most students miss out on while applying for Masters is that they do not properly research the programs to which they are applying. This is very important in a field like Data Science because the curriculum widely varies from university to university. You need to look at the curriculum and decide if it suits the type of work that you wish to do after Masters. However, if you are not super-sure about which particular field you wish to pursue in Data Science or AI/ML, then there are certain programs that offer a good mix of different courses. You should apply to these instead.
6. What are the factors you think are most important to secure a good position in the ML and Data Analytics field? What do you feel made you stand out?
I think as with any other thing in life, the most important thing is that one has to enjoy using her/his quantitative problem-solving skills to create business. There are some skills that are required to succeed in the job but as long as the passion is there, other skills can follow. But some of the skills that are worth honing to land interviews are basics of probability and statistics, linear algebra, mathematical models, data structures and algorithms, and some basic understanding of languages/tools like SQL, Python/R. Having basic competency in some of these skills is a necessary condition to be considered for a role but, I believe, having a deep understanding of the basics and knowing how to map a business/product problem to chunks of solvable data science problems, can really make someone stand out.
7. What other fields did you explore before settling for Machine learning?
Going into my Masters program at Carnegie Mellon, I knew I wanted to be in a quantitative field as I realized that I am good in that but didn’t know what exact field or role I wanted to pursue. That’s also a reason why I chose Carnegie Mellon’s program over others as it was a very multi-disciplinary program with courses in computer science, machine learning, statistics, econometrics, financial engineering, and digital marketing. Plus, the program was elective heavy and you could design your own program to tailor it to the role you were looking for. Before coming into the program, I was interested in roles like hardcore machine learning, product analytics, quant researcher with a bank/hedge fund and HFT. But during the course of the program, I tried to get more data points for each domain by taking a few electives and reaching out to alumni who were working in those areas. Due to this exercise, I became more inclined towards Machine Learning and focussed more on it towards the later half of the program
8. Many of the students learn ML and Deep Learning from online courses. How much did it help in your industrial applications and what is your review of MOOCs?
I would rate MOOCs very highly. There are a lot of great online courses where one can not only learn the basics of ML/DL, but also get hands-on practice on open-source datasets. These courses are designed by some of the experts in the field, both in academics and industry and thus they often connect with the problems faced by the industry today, which tend to evolve rapidly. I did the deep learning specialization by Andrew NG on Coursera and it helped me understand deep learning, right from the equations of backpropagation to building computer vision models using TensorFlow. When I did the course, I was a DS at uber and I was able to use the DL know-how I acquired because of this course, to create additional business impact by deploying deep learning-based NLP models. I would highly recommend MOOCs for anyone looking to learn something new or build on their skills, be it in ML or otherwise.
9. Do you think CP(competitive programming) is important even if someone is interested only in the ML field?
If you are asking if it is a must for data science and ML roles, I would say no. I actually never took part in any competitive programming competitions/events. For ML, a functional understanding of the basics of data structures and algorithms is required and that can be built through courses in your program and by practicing problems on some online code editors like LeetCode. I did around 400–500 problems from LeetCode which helped me during my interviews. But that being said, CP can definitely help with improving programming skills and could be fun too. I should actually try it out sometime:).
10. What are your long-term career goals? Do you plan to come back to India in the future?
Well, what I have learnt is that careers are not linear and thus I never think about long-term career goals. When I was doing mechanical engineering, I had no idea I would be doing Machine Learning one day. The only thing I want is to be in a space where I am solving (hopefully hard) problems, hopefully for a large number of people, and to have fun and continuously learn while doing that. And yes, even though I haven’t planned for it and I don’t want to constraint myself with it in the medium term, but I would eventually love to move back to India
11. As someone who has worked in big product-based tech companies in Data Science and ML space, where do you think the field of ML or DL is heading in the future? Specifically, what do you think would be the trajectory of NLP?
I think it is going to become more and more part of our lives. A few years back, we did not have products that could be completely based on NLP. But today we have products like Alexa, Google Home that are so good that they can understand different accents, different languages. NLP combined with Computer Vision is going to be really powerful. You go to some supermarket and they already know what coffee you drink at 8 AM in the morning. Your basic order is already populated. This is going to replace jobs like cashier, sales people in supermarkets. But wouldn’t that lead to a rise in unemployment?
I don’t think so. The change would be gradual and the kind of jobs that you have in the society would also evolve. For instance, the job of an Uber driver didn’t exist 15 years ago. I think these jobs would evolve to a point where people are doing high-quality work and at the same time working less due to increased productivity.
12. Which hostel were you part of? Can you recall your best memory from your hostel days?
I was a proud resident of the Umiam hostel. We were the second batch that was admitted into the hostel so all the facilities were pretty new and better than all the other hostels (humble brag). I spent way more time there than any other part of the campus, including lecture halls (don’t tell my professors), so most of my amazing memories of the four years are from the hostel. So, it will be hard to choose one. The first year in the hotel was interesting as me and my friend were the only two freshers in the whole wing of second yearites. You can only imagine the number of “social interactions” we had with our seniors. I chose that particular room due to its closeness to the canteen, not realizing that all the seniors also had the same idea (:P). But it turned out to be great as I made some of the best memories and great friends because of it. Being close to the canteen also helped me become good friends with the vendors. Throughout the four years we spent numerous hours talking about random stuff, exchanging gifts on festivals etc. but probably the sweetest memory is the one where the vendor brought fresh fruits from his village and gave it to me when I was leaving the hostel after four years. There are so many memories flashing in front of my eyes right now and I really miss the four years I spent there. I would give anything to be able to relive those years.
13. Which hostel used to dominate the GC back in those days?
I actually don’t remember which hostel won the GC each year. I guess I need to eat more almonds to improve my memory. But, there were a couple of hostels who were always the top contenders: Manas, Kameng, and Kapili would always dominate everything. Barak (for obvious reasons) and our hostel, let’s just say, were way lower in the rankings than we would have liked.
14. While in IITG, were you part of any clubs or fest teams?
I did participate in some of the clubs but looking back at the sheer number of clubs and facilities we have in our university, I feel like I should have taken part in more. I was part of the Mechanical Engineers Student Association (MESA) for a while and we tried to organize some student competitions related to what we were learning in our program and also helped the batchmates with internship search. Apart from that, I did play a small role in organizing the logistics of Techniche, the annual tech fest, in my first year. That one may have been against my free will and me being in the wing with a lot of seniors might have a role to play in it (just kidding).
15. What were your favorite hangout spots back in your college days?
It has to be the Serpentine lake and the hill near the guest house. It’s like a little oasis in the middle of the campus where we could go and relax. But if you consider the total time spent metric, and exclude the hostel room, canteen and academic area, the terrace of Umiam hostel might win as my favorite hangout spot, especially when the weather is great. I hope to visit the campus sometime soon with my friends.
16. How difficult was it moving to a new country for studies and then working there full time for the past few years?
Naturally, moving to a new country can be a little challenging and fun at the same time. I saw significant differences in the way things are done: right from the way professors run a course to expectations in a workplace. Adjusting can be a little uncomfortable in the beginning but I started enjoying it after a while. I started meeting so many interesting people from so many different backgrounds that I would have never met had I not moved to the US. During the process, I learnt to enjoy and celebrate the differences and strangely I started to understand my background and myself more. Overall, despite the initial challenges, I have been enjoying my time here.
17. How often do you come to India? Have you been to the campus after your graduation?
So far, I have made sure that I spend at least 4–5 weeks every year in India no matter the situation. Some years it’s been more than that. At the end of the day, the thing that matters the most is the family. Even though I have made a lot of friends here, all my family is in India and for some reason, they don’t like visiting the US. So, I am left with no option. Unfortunately, I haven’t been to the campus after graduation. I can only hope I get a chance to visit soon with my friends.
18. If you had to start a startup, what would it be? Do you ever plan on this?
I haven’t planned for it yet but this is something I do think about. I love using whatever limited skills I have to solve real problems and there are multiple ways to do it, startup (either starting your own or joining an early-stage company) is one of them. Another option is working on new initiatives/products that are like start-ups in a big organization but there are pros and cons for both. I have been doing the latter so far but if I find a problem that I am really passionate about, and the people that I admire, I would definitely go for it.
19. What were your hobbies and do you still take out time for them after your work hours?
My hobbies have actually changed a lot over the years and yes, whatever my hobbies are, I always take time out for them. It’s good to have hobbies outside of work to which you can commit time. It keeps you going. In college, I was a big movie buff. I spent most of the time checking out thriller movies and shows. Apart from that, I used to play cricket and football and was part of a few inter-hostel tournaments. There was a phase where I was into driving a lot, like go-karting, renting old sports cars, etc. For the last two years, I have also been playing tennis and try to play at least twice every week. I also like to travel and my favorite travel destination is Europe.
