Hi. Could anyone speak about how data science is at capital one? Even if you aren’t in the role, have you heard anything about it? I am a current master’s student who is very interested in their data science program, and would greatly appreciate any insight. I would also appreciate a referral! I have worked previously in research, insurance, and have been a data hackathon finalist. TC - 0
I'm a data scientist at Capital One and we do work on interesting problems. I hardly ever use Excel and most of my work involves machine learning and engineering. I think it mostly depends on the team.
I'm a c1 DS as well. I've been around lots of interesting problems and have always had to of the line open source tools and cloud infrastructure available. I've touched Excel maybe once for a minor report plot. I've heard of BA's using it heavily, but never seen a DS do modeling in Excel
The DS culture is pretty great - lots of cool projects going on and good learning opportunities. DS holds pretty decent respect/value with the business units as well. It's better than most Banks here, but even a perfect tech culture at a bank will be fundamentally slower and more restricted than a tech company because of all the anti-bias regulation. But the thorough validation work required can be interesting in it's own right to some extent.
Excel guy sounds like an anomaly. All the DS I know are doing albeit basic machine learning. Random forests everywhere. Deep learning is only being touched in small pockets of the company as far as I can tell
Deep Learning is too advanced for Capital One
So first I would agree that from the ds teams I’ve seen here, they have some pretty interesting algorithmic problems compared to the rest of the company. You need a master’s degree to do data science so you are fine there. The issue is: Capital One does not have enough data to apply interesting conclusions or solve problems. We do not operate on TBs of data. The majority of data we have is probably uncleaned/scattered everywhere such that it cannot be used by data scientists. Whatever learnings ds are generating I have not seen used in any of our products except maybe fraud, but I imagine they are using external fraud models or its very basic
Also this is a bank not a tech company. So data scientists generate the models in python. Once they finish their work, they need to convert it into scala or java so it can run fast. But the ds do not know how to do that so it gets handed to another team who does and they convert it. With the politics and the extra work and meetings and general regulations, you will see your model likely being published in 3-4 months at best.
Depends on the team & platforms you're working with - some LOBs have tools that package the python model & let the DS teams deploy to prod themselves.
Yeah I think Capital One definitely recognizes the deployment process as a problem and has tried to implement solutions. Only problem is that again, they don’t put priority on this when they can build features / deal with new regulations
Wow. Thank you all for the valuable input. The fact that you guys have interesting projects to work on is probably what I value most as someone looking for an ds opportunity. From an interview prep perspective, what do you guys stress on working on in my free time? The reviews/questions available on Glassdoor provide a lot of insight and I’ve looked through that; anything else you guys suggest?
From my experience, Capital One’s interview process is a joke. The questions are insanely easy. YMMV
Tech Industry
Yesterday
3403
ByteDance is officially fucked
AMA
Yesterday
2018
I’m a professional coaster AMA
Ask Blinders
Yesterday
1111
Tipping culture is really getting out of control! Waiter gave me ‘a look’ because I tipped her 10% for ‘BAD service!’
Tech Industry
Yesterday
2348
TESLA UP 14% AFTER HOURS 🎉🎉🎉🎉
Tech Industry
2h
557
I wish I were East Asian instead of Vietnamese (Southeast Asian)
My friend who works in data science there works on visualizing time series in Excel after pulling data from an oracle R11 DWH. While this is a good start for a large bank, the skills one develops here won't translate well into the tech industry
Interesting, so nothing machine learning related?