Now Reading
5 Reasons Why… 60 Days of Brainstorming, Coding and Testing Was Worth It At YES Datathon

5 Reasons Why… 60 Days of Brainstorming, Coding and Testing Was Worth It At YES Datathon


 



What’s common between data scientists from GE, Uber, Boeing, Walmart Labs, Capillary Tech, Big Basket – well they are super busy, are great at the most sought after job in the industry and they spent their Christmas eve weekend (Dec 22/23) at YES Datathon hosted by YES BANK.  Surprising! Even we were when we considered that over the last year hackathons have becoming quite commonplace, especially in the startup hub – Bangalore.

So what drew them, two things– first, the fact that it was a showcase of models developed over 60 days – a decently long period to engage and work with hackathon participants, we wondered what exactly were these models being developed; and second, the volumes of data and variety of use cases and algorithms being explored – from transactional social communities to predictive customer service – Data the opium for Data Scientists drew them all.

Our interested piqued, we wanted to find out more, and we joined YES BANK’s Datathon showcase in Bengaluru on December 22 and 23.  The setting was familiar on Day 1, a large well decked co-working place thronging with 200+ data scientists/developers hard at work on their laptops and thronging the coffee bar.

Appearances, however can be misleading and once we had spoken to some of the teams as well as the bank’s exuberant data analytics and strategy team, we got a taste of how this was a really different quest that these data teams and the bank had set out on.  The fact that the finalists were chosen from 6000+ applicants through 5 Machine Learning challenges added more heft to the challenge.

See Also
cypher 2019 experience

The participants, most of whom were working professionals with cross-sectoral experience as well as some really bright young data science students, had identified use cases in the banking sector almost 60 days back, post which they had worked on terabytes of anonymized data sets , and developed models which the bank and the teams were going to implement over the next few months. So this wasn’t just looking at innovation, it was more implementing innovation with a ecosystem/community approach. So what made it worth for the participants to work over such a significant duration beyond their work hours/classes. Here’s some cues we picked up from them.

  1. It’s all About the Algorithms: Staying curious is key for any data scientists, and our chats with some of the teams whether that was Finance Data Dons , a team of TCS data scientists (Karthikeyan, Chitra, Ramnathan) with more than 20 years of experience between them or Reverse Atlas a team of students from VIT, Vellore it was this desire for exploratory data analysis that drove them. With mountains of data at disposal, the teams had the chance to go deep into the mathematics of algorithms , explore and apply them on the data to create working models. For instance, a team of young data scientists all the way from Sri Lanka used this opportunity to apply a couple of research papers they had read on Shap Values and Shapley’s method. Another fact which seemed to resonate with participants was while most were non-bankers but all were customers and had taken up challenges which they felt their bank should address
  2. Then there is data: As any data scientist will tell you innovations in algorithms are great, but first you have to make sense of the data. The very fact that there was a ton of data, albeit completely anonymized and non-representative made this an even stiffer challenge, which made for the even harder effort put in by the teams. Some of them like Team Billa, Ajay and Mohsin from Walmart Labs and Happiest Minds even worked out innovative models to first tag and meaningfully reduce the data before building their models. In fact Rajat Kanwar Gupta ,Head, Business Analytics, YES BANK mentions that his team was constantly working 9 pm – 2 am shifts with the participants to ensure that they could make sense of the data sets and also had the correct data sets that their models demanded
  3. The Environment Matters: Something that coders love about hackathons is the freedom to use their own tools , environments, libraries and the works but as some of us understand as a data scientist, engineer, full stack developer you often work in defined environments given the safety and data security standards of any organization –startup or corporate. YES Datathon was a different challenge in this regard as well. The bank had a dedicated set of Hadoop engineers working on setting up an environment to mimic their own with AWS and Cloudera tools for all the top 50 teams. This was unique and the participants agreed that while it took a bit of a time to adapt, soon with the almost round the clock support and guidance from YES BANK’s teams they had absolutely nailed that as well.
  4. Cross Sectoral Mentorship: When we looked at the Datathon website, one question rattled us – given the aim of developing working models on banking use cases, how relevant would a cross sectoral mentor pool be. Our questions were answered by the participants – one of the teams told us how they were using a technique similar to Google adwords to cluster transactions, while another was using an algorithm used by streaming sites to classify your favourite movies. With experts from Capillary Tech, Uber, GE among others the participants were exposed a variety of problem solving techniques which not only enriched their experienced and approaches but also expanded the bank’s horizons, admits Rajat in our discussion.
  5. . the recognition matters as well: Well, like any other competition there were handsome rewards for all the winners – a difference being all top 10 were awarded cash prizes , it’s something else that stood out. Our conversations with several leaders from the bank across Strategy, marketing and data teams revealed that they truly believed that not all innovation could come from within and this was also a drive to learn and unlearn their perception of data innovations. Countless times, the teams had detailed conversations and demo-checks with the participants which also kept the tempo up all the time.

By the time we wrapped we realized that the heady cocktail of these 5 had set a recipe of a great ecosystem connect and community led collaborative innovation.  We will certainly be looking out for other organizations to join the fray, and also at YES BANK’s promise to further build on this with year long initiatives.


Enjoyed this story? Join our Telegram group. And be part of an engaging community.

Provide your comments below

comments

What's Your Reaction?
Excited
0
Happy
0
In Love
0
Not Sure
0
Silly
0
Scroll To Top