When Shekhar Devm Upadhyay summarizes his summer internship at SEB, this is how it sounds: “I like this place and I never expected corporate life to be so open and inclusive. When you feel you are part of the decision-making process and can challenge the way things are done, you will also be more invested in the success of your work.”
But let’s start by going way back.
Growing up in India, Shekhar’s childhood was deeply rooted in Indian culture, philosophy and tradition. A passion for singing and playing instruments made him consider a career within music, before deciding to keep it as a hobby and pursue his interest in natural science instead. After gaining a broad understanding of several subject areas, he chose machine learning as his main focus.
Shekhar soon came across the MSc. Machine Learning program at KTH, decided to apply, and was admitted along with the KTH-India Scholarship. Off to Sweden it was!
The studies went well and during the second last year of the programme, in the beginning of 2023, Shekhar won employer branding firm Universum’s competition Tech Talent of the Year. As part of the prize, he was awarded an eight-week summer internship at SEB, which is now just about to finish.
“It’s been great. All in all, I’ve been at four different departments – three in Stockholm and one in Vilnius. What really stood out to me was how easy it was to feel part of the different teams. I’m also very happy that I’ve been able to work on my own project, getting useful experience in applying my knowledge to real-world cases”, Shekhar says.
Shekhar’s own project, yes. To someone completely uninitiated this may seem, ahem, a little complex.
Let’s get technical
What Shekar has been working on is improving extractive Q&A-systems, designed to find correct answers from large bodies of text. By employing adversarial machine learning – which basically entails trying to trick the systems into identifying the wrong set of words as the answer and having them learn from such mistakes – he has been able to strengthen the systems’ overall ability.
“Something like this could save employees at the bank a lot of time, by filtering out the relevant information from the vast stores of data available and recommending a course of action. We could take decisions in days instead of weeks”, Shekhar says.
“There's a good amount of initiative in the different teams within the bank to become more data driven. The perhaps most important step in this process is to showcase what AI can do for the business and get all stakeholders on board. That way, we can slowly grow the data science capabilities in different parts of the organisation."
Sounds like you’re on a mission – will you be coming back to SEB, by any chance?
“Fact is, I will continue working with SEB through the autumn, in addition to my studies at KTH. As for the future, there are a lot of possibilities to explore, both personally and for SEB. I feel we're on a path towards something really interesting.”