J.P. Morgan is one of the most advanced banks when it comes to data science and machine learning. It hired in Geoffrey Zweig from Microsoft in February 2017 as head of machine learning. It’s actually launched a market-making product (LOXM) based on machine learning and it recently promoted Samik Chandarana, a former credit trader, to head its data science and analytics (effectively its machine learning) strategies.
Chandarana hasn’t actually started his new job yet – he’s still a trader, but he’ll be starting it soon and in an interview posted on J.P. Morgan’s Youtube channel, he expresses various opinions about what it will entail. The bottom line, as Saeed Amen explained in a recent blog, is that machine learning and data science jobs in investment banks aren’t necessarily as exciting as they seem.
Chandarana alludes to the fact that far too many machine learning projects in investment banks come to nothing at all. The record of, “productionization of innovation” at banks is, “challenging,” says Chandarana. In other words, banks innovate but nothing comes of it.
To overcome this, Chandarana says people working in his area need first of all to, “get the trust of the underlying customer,” and that initially these customers will be, “internal.” Like other exponents of machine learning, Chandarana reiterates the notion that machine learning will eliminate mundane tasks and free human beings to pursue higher value work. However (and he doesn’t say this), it’s clear that internal customers might be resistant to machine learning because they think it will steal their jobs.
Despite internal clients’ skepticism, Chandarana says the aim is to make machine learning and analytics part of people’s daily lives at J.P. Morgan.
However, this isn’t going to happen quickly: a process must be followed. First of all, Chandarana says analytics teams need to collect a set of, “problem statements,” describing exactly what internal clients need. Then it’s necessary to see if there’s enough clean data to solve these problem statements (and to clean it or find more data if there’s not). Only then can the team start building solutions, and when it does it mustn’t get to carried away. Twice, Chandarana says they need to keep things simple: if regression analysis is all that’s required, regression analysis is all that will be used. The implication is that machine learning and analytics teams like to over-complicate issues.
Chandarana also says he’s more interested in hiring people who “know how to work with customers” and are “like-minded” than pure “book-smart” talent. However, he acknowledges that J.P.M’s analytics team will need to partner with academics and fintech firms (which might have more book-smart) people if it wants to remain up to date.
Lastly, Chandarana seems bemused to find himself heading J.P.M’s analytics business. He describes himself as an, “accidental banker,” and says he aspired to be a techno DJ whilst at university. He collects vinyl records and likes physical books. If you want to present yourself as a “like-minded” soul during a J.P. Morgan analytics interview, these are probably good things to know.
Have a confidential story, tip, or comment you’d like to share? Contact: email@example.com
Bear with us if you leave a comment at the bottom of this article: all our comments are moderated by actual human beings. Sometimes these humans might be asleep, or away from their desks, so it may take a while for your comment to appear. Eventually it will – unless it’s offensive or libelous (in which case it won’t.)