Hi guys! I’m Tezan Sahu, a senior undergraduate in the Department of Mechanical Engineering, pursuing a Minor in Computer Science & Engineering. I Sing. I Paint. I Code. In this rather longish post, I present glimpses of the journey that led me to a Data & Applied Science Internship at Microsoft (as well as something pretty cool that I had not expected, but ended up doing) along with some anecdotes. I hope you all enjoy it and stick around till the end.
That’s how it all began…
The journey started much before the Internship Season actually kicked off. As an Ex-Convener of WnCC, I was a part of the “tty0” group comprising all the members & ex-members of the club. It was in June 2019, when I was interning at WandX Pvt. Ltd., that one of our seniors floated a form on that group asking for people interested in the Microsoft Early Intern Connect Program so that Microsoft recruiters could get in touch with potential intern candidates early on. Fully understanding that most of the dream companies for me (like Microsoft, Google, Facebook, etc) do not open directly for Mechanical Engineering students, I decided to grab this opportunity. Soon, I received a call from one of the University Recruiting (UR) Team members who guided me about preparing for the coding test & interviews in general.
In August, as expected, Microsoft IAFs opened up only for CS & Electrical Undergraduates. I (and some other students who had applied for the program) reached out to the UR Team, who finally made arrangements to allow us to take the tests. Since I had applied for both Software Engineer (SWE) and Machine Learning (ML) positions, I had 2 consecutive 1-hour tests scheduled that evening. A few days later, I was overjoyed to see myself shortlisted for the ML Intern Interview.
That weekend, I had to stay in the institute (which was an unusual sight for all my friends because I belong to the ghati junta who take every opportunity they can to go home on weekends). Saturday evening saw us writing a pen-paper coding test with 2 DSA related questions. On Sunday (Tech Day 1), each candidate was supposed to go through 3 rounds of interviews. I had 2 of mine before lunch, which involved questions on probability, stats, query-search and recommender-systems related stuff. I was happy not to be grilled on DSA questions :P! Later, after waiting for the 3rd round for over 4 hours, I was simply told that my “Interview process had ended”. Totally dejected, I went to sleep early that night. Hardly had I slipped into a deep sleep that my roomie yelled! My phone was switched off, so another friend told him to wake me up to congratulate me on my selection as an ML Intern at Microsoft. That was a dream come true for me!
The COVID Effects and a Surprise!
Cut to March, when COVID-19 had started raising its head and I was hearing about cases where my friends’ internships were getting cancelled. While talking to a WnCC Junior, I rediscovered my interest in the Google Summer of Code (fondly known as GSoC) lurking within me since the first year. I decided to prepare a proposal (more like a fallback option), although I knew it was pretty late & I hadn’t contributed much to the open-source organization.
With the announcement of the lockdown in April, almost all my hopes about the internship were shattered. The email from Microsoft’s UR Team stating that our internships would be honoured was a sigh of relief, although I was too sceptical to be happy about it then. Soon, we were briefed about the 8-week Virtual Internship Program by Microsoft.
May 4 was a day that I would hardly forget. To my sheer delight, I had been accepted for GSoC with the PEcAn Project, to develop their API & R package over the course of 12 weeks starting from June. It dawned upon me that the month of June will definitely be a challenging one, so I began preparing accordingly.
The Internship Commences…
The first day of my Virtual Internship at Microsoft was full of onboarding sessions that helped me set up my virtual development environment. I was a part of the Search Technology Centre India (STCI), commonly known as “Bing” team. The next day, several projects were pitched to us by our to-be mentors & managers & we were asked to select our preferences. Things here started a bit weirdly as I was not allotted any of my 6 preferences. I was assigned to the Core Relevance Team, with a project titled “Analytics for Core Ranking Offline Experiments”. [I was as confused that time as you guys are reading all that jargon!] After having a chat with my mentor & manager, I got a vague idea about what I was expected to do, and we chalked out some deliverables that were expected out of this internship. The project was not very flashy, with the core idea being to build a pipeline that would allow ranking engineers in Bing to draw useful insights from the experiments that they try out with different search results ranking techniques. At that point, even I was unaware of the potential impact of this project.
Ramping up in the first week was a tough task! After all, you can’t expect someone to build a debugger for a system unless he thoroughly understands the processes. So it was imperative to understand the Bing Ranking stack and how engineers train new techniques. With consistent support and resources from my mentor (and a couple of other teammates), I was able to get some grip on the stack while being able to accomplish my first subtask by the end of week 2. There were several Microsoft internal tools that I got a chance to work with as a part of this project and eventually, I started enjoying stuff that I was building. Also, every Friday evening, our team had a “Fun Time” where we would all play games & chat about random stuff. It was really refreshing!
Meanwhile, on the other front, I initiated discussions with my GSoC mentor and asked him if we could start early considering that in the month of June, my attention would be diverted. He agreed enthusiastically and so I started my GSoC project parallelly as well. Being in timezones almost 10.5 hours apart, the odds were in my favour as I could work on my internship project during the day while managing the GSoC project post 8 PM daily, andon weekends. Frequent 2 AM chats with some friends were the most energizing moments after such long working hours!
Due to the virtual nature of the internship, networking opportunities were limited, but I tried to make the most of my time & connected with people from various teams across Microsoft. Everyone was really welcoming, and by the end of week 3, I had a side project going on, with a mentor from the Related Searches team, based on query specificity classification.
Striking the Balance!
The first two weeks of June were really eventful, where I chanced upon attending several talks & tutorials at the Machine Learning & Data Science (MLADS) Virtual Conference. Since the timings were according to PDT, I managed a total of 6 night outs during those 2 weeks (which is probably more than the total number of night outs that I have had till date in the institute :P). As work on my internship project progressed, I started understanding the kind of impact that this project could create. Regular scrums & syncs with the team allowed me to derive ideas on how to structure this seemingly fuzzy project idea. Slowly, I was able to come up with relevant analytics & inferences that would reduce debugging time from over a week to just a few hours.
A major highlight of this internship was the Intern Learning Week, where we interns were exposed to a series of talks focussed towards Microsoft’s Culture & Technology. This concluded with the Intern Talent Night, where ~15 interns (including me) performed virtually (singing, dance, playing an instrument, poetry & lots more!). The event was truly one of its kind & was attended by many Microsoft interns & employees. As an aside, this event sparked off my long-standing desire to create a YouTube Music Channel & I did start one amidst this internship.
As the internship approached its final stages, we actually saw the project take good shape & by the end of the 7th week, we were almost ready with the end to end framework, all set to test different experiments. The last weekend was spent in preparing the final presentations. To my delight, the presentation went great & the higher management did talk about sharing this project across different teams. During the last week, my major focus was on the knowledge transfer for the internship project & building the DL model for the side project. I also got a chance to author & submit a short paper about my internship project for the MLADS Synapse Conference (the Asia-Pacific counterpart of MLADS that happened in early June).
All’s Well that Ends Well…
A virtual internship in the given setting was surely a challenging yet fulfilling experience. The joy was trebled with the selection for GSoC 2020 & the opportunity to work on a side project of my interest. I am really happy that I have been able to deliver the intended impact through my internship project, while also having reached the proposed milestones before the 1st GSoC Review. Moreover, I also plan to publish my work on the side project with my mentor soon.
A key takeaway from this experience for me is the fact that no one or no situation can deter you from performing well & creating an impact as long as you have the will to fight it out till the end & also have the right set of people around you. It has definitely given me an upper hand in preparing to embrace the “new normal” that we are all heading towards.
Well, you have finally made it to the end of this long post! Thanks for sticking around. I hope you guys are able to derive some value out of my experiences. All the best, & stay safe!