I am Eeshaan Jain, a 4th-year undergraduate in the Department of Electrical Engineering & I am pursuing my minor in the C-MInDS department. I will share my internship experience at AWL Inc, Japan, during my third-year summer. This blog post would essentially consist of how I prepared for the internship season and my experience during my internship.
Pre-InterviewSince the second semester, I got interested in Artificial Intelligence and Machine Learning and went ahead and completed many courses online and offline. The more I delved into the concepts, the more I realized this is what resonates with me. With this newborn interest, I spent the next year doing various projects and courses (of other universities, too) and reading blogs, papers, and books. With so much enthusiasm, I had already decided that I wanted to pursue this further - not only in internships but also in courses, projects, and future work! Soon enough, the year passed, and on came the 2nd-year summer, where the dreaded internship preparation started. In the back of my mind, I knew I wanted to target either a company ML RnD or a university research role. This thought made me focus only on a subset of opportunities and made me selective in preparation (P.S. that it is anyways selective as I am in D.D.). I spent time browsing the InstiApp for options that came in the previous years and talked to people selected there to get an idea of what I should prepare for. I soon understood that there are three significant components I need to work on - building a good resume (and being thorough with it), preparing for standard ML/DS questions (along with knowing what the target company/univ dealt with), and finally DSA (uh-oh).
It took a considerable amount of time to build the resume as I had to make many revisions and changes after consulting friends and seniors (relatable for everyone - the final file was called "2_page_final_ab_aur_nahi_hora (3).pdf" 😛). But after dreaded weeks of just making the resume, I finally went on to revising ML/DS and started to get anxious about DSA as I had never touched that formally. The former consisted of scrolling websites, cheat sheets, blogs, and videos of common preparation questions and analyzing all projects in my resume (going over the project idea from scratch, seeing rooms for improvement, and reading nearby theory). Since there is so much to do, I planned it accordingly to do it over the next couple of weeks. The next part was DSA - all I did was go over all data structures & algorithms on blogs/youtube videos and sit for hours a day on LeetCode/HackerRank etc. I wouldn't say what I followed was good, but it worked somewhat for me in my restricted time.
Internship SeasonAfter this preparation ordeal (and more resume revisions), the IAFs started coming on the placement cell's website. As each day passed, I saw that either there was no ML RnD role in the company, or it was not open to D.D.s. As this continued, it dawned upon me that I should instead start university apping.
Soon after that, AWL Inc, Japan, opened up its IAF in the RnD profile, and finally, there was some hope. Not only was it the role I wanted to work in, but it was also in Japan - one of the places I had wanted to visit for a long time! The selection process was two-stage: an online test followed by a technical interview. The online test consisted of questions on DSA, ML, and Docker. After completing the online test, the interview shortlist was up in a couple of days, and it was exciting to see my name up there. After reading blogs (just as you are) - I learned that most companies from Japan ask most of the questions about yourself and your resume. Hence, I thoroughly prepared topics on my resume, and finally, the day of the interview came.
The interview was so much different than what I anticipated. The panel had people from Japan and India who were incredibly cool. They first made sure I was comfortable and told me to relax. Asked for my introduction and went on to some general questions about me. The next phase consisted of questions based on my resume. Since the interview was only 30 minutes long, they asked about two projects I had put in. It was clear that they had reviewed my resume and chose projects related to the field to ask for. Not only did they ask about the work done, but also about room for improvement in the project, along with technical questions on concepts used. The interview was interactive, and it felt like a conversation!
I felt that my interview went really well, and the waiting game for the results started. Surprisingly, in a few hours, the results got released, and I was elated to see my name on the final selection list! Finally, the grueling procedures for the internship season ended for me, and I was glad my preparation paid off.
COVID-19During the internship season, COVID-19 was at an all-time peak. All places were in lockdown, and most companies were working from home. The new circumstance caused a bit of unrest. If the situation didn't improve, the internship would resort to an online one, and all my plans for visiting places and experiencing the Japanese culture would be destroyed. Soon enough, the COVID-19 situation improved, and the world hit an all-time low, with many businesses opening. The improvement in the situation gave a bit of hope, but that didn't last long as we got an email saying that the company is trying for an offline internship, but the COVID-19 situation is not good in Japan. I checked, and that seemed like the case - Japan was soaring at 50,000+ cases a day! Alas, finally we got the mail saying the internship is online. That sucked, but there is not much one can do when the entire country is at the pandemic's peak (P.S. - at the time of writing this, too, Japan is at 100,000+ cases per day).
Internship ExperienceWhen the internship started, I was surprised by how calm, happy and well-natured everyone was. Our introduction session with the entire A.I. team and direct supervisors ended on Day 1, and they told us about our tasks during the internship. Since AWL Inc, Japan, is a startup, a lot of our work was going to directly affect the company's work; hence it was super exciting and a motivation factor to work hard during the internship. It also added discipline to my daily schedule - since the work needed to start at 7 AM each morning (to sync up with Japan time).
My work revolved around accelerating neural network inference - speeding up the inference time based on hardware intrinsics of devices. I wasn't familiar with this, as I was always used to running models on GPUs and CPUs and hardly ever cared about timing. But here, I realized the importance of inference times from the perspective of deploying models - especially at remote locations. The added challenge was that not only did we have to leverage our regular CPUs, but we also had to work with low-power devices and remote GPUs. The task was enjoyable and, in the end, exposed me to a new ML/AI deployment domain, which I had never focused on before.
Overall, the internship exposed me to a more practical side of ML and made me realize how much effort is required to deploy so many models and how much more there is to learn!
FinallyThank you so much for reading through this blog I wrote - P.S. I am not much of a writer. If you have any questions or queries, feel free to contact me 😄