Hey everyone! I am Kunind Sahu, a BTech thirdie (to-be-fourthie) in the MEMS department. I’m super enthused by applied ML - especially Graph Machine Learning. I also love to play video games and watch thriller/horror shows. Yes! I’m back to bore you with yet another long write-up, this time about my summer experience as an Analyst Intern at American Express in their Credit Fraud and Risk Wing. So, without further ado, let’s get into it.

Points 1-5 talk about my experience in the internship season and AmEx’s selection process. Points 6-8 talk about my actual internship experience.

  1. Internship Season Profile Selection

    I kinda had an idea of what I wanted to do - I wanted to go for roles which involved Machine Learning - essentially the analytics domain (although not the entirety of it). I was sure that I wasn’t interested in applying for roles along the lines of ‘business analyst’ because they were non-technical in nature.

    The basic idea was that I had done a second year summer research internship - where I was exposed to how Machine Learning was being employed in an academic setting. This time around, I wanted to experience the corporate world and how Machine Learning was being used in the industry - maybe in some kind of industrial RnD or some applied Machine Learning.

    Although I did not enjoy software development much, I nevertheless decided on applying for some software development roles as well. I had done a course on Options Pricing over the summers through Finsearch, organized by Finance Club - that got me all interested in the field of quantitative finance as well, so I applied to a few quant finance roles too. In summary, the profiles I decided to go for were - Analytics (ML), Software, Quant Finance.

  2. Preparation for the Internship Season

    I have written in detail about my internship season preparation right here.

  3. Internship Season

    The internship season was nerve wracking to say the least. As I’d mentioned above, I wanted to work with ML in the corporate setting - either an RnD role or some good application based ML role. Unfortunately, mostly, none of the Day 1 companies were open for my department (only open for CS, Elec and sometimes BS Math) - which, it goes without saying, was quite demotivating. Apart from one company (Optiver), Day 1 went by without much activity with respect to shortlists/selections.

    The American Express IAF had opened just after the Day 1 companies announced their results. Their job profile fit well with my expectations of my third year summer internship. American Express was hiring in their Credit Fraud and Risk (CFR) division for the upcoming summer. The CFR team is responsible for developing state-of-the-art Fraud Detection models to prevent credit card fraud and catch possible defaulting clients.The fact that I had quite recently done a course project on Credit Card Fraud detection made me all the more excited about this role. Interesting team? Check. Good work? Check. Great company? Check. The IAF was immediately signed ;)

  4. Background about AmEx

    AmEx is a financial services company, which issues its own credit cards and corporate cards - which are invite-only. It usually tends to operate in the premium high spending segments. In this way, it is different from Visa and Mastercard.

    Any credit card transaction has three entities associated with it. You have the consumer, who’s buying a commodity with their credit card, you have the merchant who accepts the card and sells the commodity, and then you have the technology network over which this transaction takes place (Visa, Mastercard or AmEx). Let’s assume the customer’s bank is HDFC, the transaction network is Visa and the Merchant’s bank is SBI.

    Visa and Mastercard don’t issue their own credit cards - they are issued by the respective banks. In our above example - the customer’s data is held by HDFC Bank, the transaction information (and metadata) is held by Visa and the merchant’s information is held by SBI. No single entity has access to the complete data about the transaction here.

    This is where AmEx comes in. AmEx issues its own cards - thus AmEx has complete information about any of its credit card transactions - information about the customer, about the transaction and that of the merchant. This forms a closed loop, which AmEx leverages in its ML models to make them state-of-the-art. Having access to the complete data enables them to better capture fraud and mitigate potential risks. Additionally, it was the very first company to deploy an ML based fraud detection model in service (2014).

  5. AmEx Selection Process

    Thankfully, there were no DSA rounds in the AmEx selection process The selection process involved:

    a. Resume Shortlisting

    b. ML Based Test (Those who did well were shortlisted for the interviews) Which comprised very easy MCQ questions on Machine Learning (basic ML knowledge was required to ace the test)

    c. Interview (2 Rounds)
    The interview comprised two rounds - none of them were really demarcated as Technical / HR. Both kinds of questions were asked in both rounds. The major focus of the interviews was my resume, specifically, the projects I did, especially the one on credit card fraud detection. We had worked quite extensively (tried many techniques, quite a few of which failed) - which made for quite an amazing discussion
    I was pretty familiar with the mathematical details and workings of all the algorithms I had implemented in my projects - thus I tried to steer the interview in that direction. I did go quite deep into the mathematical aspects of the algorithms I had used. Apart from the resume, a few AmEx specific questions to know if I knew about the company and two puzzles. All in all, I left the interview feeling quite confident about my performance.They had also asked me about my project preference - either Big Data & Machine Learning or Data Analysis. I chose the former

  6. The Internship Experience (Offline?!)

    I was looking forward to an offline corporate experience and getting to interact with all my co-interns from other colleges. Although, to our dismay, the internship was being conducted in an online mode. Although, internally, AmEx had this policy where employees could voluntarily choose to come to the office. Once I got to know this, I discussed it with my Manager and packed my bags and left for Gurgaon ASAP!

    My project could be considered as an applied RnD. It mainly revolved around applying Unsupervised Machine Learning, ML model explainability techniques to understand model outputs and prescriptive analysis, to a problem the company was facing.

    Oh boy, I’m so glad that I chose to go to Gurgaon and work in the office rather than working from home. AmEx owned an entire building in the One Horizon Center campus in DLF Phase 5 Gurgaon (although that was not its best office in Gurgaon. The best office, which was on the top floor in the main building of One Horizon Center - was closed :/). I got to meet my immediate and the extended team and was able to network with all of them (atleast when they did come to the office haha). AmEx had this awesome complimentary cab service which could be availed to travel to and from the office from your residence. Additionally, the workspace was quite open and there was a huge monitor on each workspace - (a dream for any programmer!).

    Talking about Gurgaon, I was kinda let down by the city - especially by a not-so-vibrant street food culture, even though Delhi was so close. But yeah, we did explore the city a lot and found some good places to eat (Sector 46!). My PG flatmates (the ones from IITB) and I did go out over the weekends and went around - especially to places in Delhi. I did have many insti friends interning in Delhi, one of whom claimed that Delhi was much much better than Mumbai (personal attack) and I was expecting that friend to show me around. Long story short - don’t expect anything of your consulting friends XD.

    I was working on the team of the very same VP who’d interviewed me. Mid-way during the internship, he had taken out the entire team managed by him - including the directors, managers and all the other levels had out for a treat! There were many such outings I had with my team. Haha, one of the perks of taking extra efforts and going to Gurgaon all the way from Mumbai and working from the office (and being a super enthusiastic intern xD).

    My team had this culture, where everyone met every Friday evening to chill out and play team games :). Codenames, Skribbl, Drawbattle, Psych – we played it all. Well, I had a suspicion that I was really good at Pictionary and Dumb Charades, but it was always hypothesized on small sample sizes. Now that I actually played with my team, it more or less checked out XD. The AmEx work culture is nice and supportive and people there are really helpful and eager to help you out. Anyway, this is pretty much how it went until the first week of July, by which time I had my end-term presentation.

    Frankly, I’m glad that I chose to go to the office and work from there - I was able to network with people (though not as much as I would have been able to if things were completely offline, but hey, it was something) - all in all it was an enthralling corporate experience - something which I’ll certainly remember.

  7. Skills Developed During the Internship

    I’d say that the internship was more about applying the skills I was already familiar with - to try to solve a problem the company was facing. Nevertheless, I did learn quite a few things - improved command line scripting, knowledge of new state of the art ML algorithms and patience (:P) when the ML algorithms are running (I was manipulating a massive real life dataset comprising 3 years’ worth of data - even with AmEx’s strong hardware, it took some time to process).

    It was nice to know AmEx’s business practices and their usage of closed loop to create awesome ML models and leveraging all that information to improve upon their services. Last but not the least, and one of the most underrated of the lot - presentation and communication skills. It was super important to present my work and findings well - the people watching my presentation would be from diverse backgrounds (technical / non-technical), many of whom were not at all familiar with what I’m doing.

  8. Final Words!

    Yeah yeah, I’m finally done with this yet another overly prolix and unnecessarily detailed write-up about yet another internship experience of mine (*sheepish smile*). On a serious note, the internship season is a good experience to have. Its also a great opportunity to learn more about your preferences and interests. Yeah, Day 1 companies are great, but it doesn’t mean that if you don’t get “Day 1”, the internship season is over for you. Not at all! Quite the contrary in fact, there are many good companies which don’t come on Day 1 - AmEx being one of them. Actively look out for such opportunities around you, which are to your liking and of course, do ask the interviewers more about the job profile and the kind of work that would possibly be allotted to you.

    Also, I can’t stress on this enough, please please please keep in constant contact with your friends; keep talking to them, prepare together, it’ll help improve both of your moods and would kinda act as a morale booster. Internship season is of course a super important period (and super stressful too) in your stay here at IITB. There are surely going to be times when you are going to feel really down and not up for it - talking to your friends would always help you out. Also, it isn’t a zero sum game, help them and let them help you - don’t compete with them, rather, walk together :). Thank you soo much for staying with me until the end of this long monologue-ish story write-up. Another shout out to Insight for giving me this opportunity to pen down my thoughts (read: blabber) about my internship. All the best! Do feel free to contact me, I’d be happy to help in whatever way I can!