How do I become a data scientist?
September 12, 2015 3:47 PM   Subscribe

How can I get an entry-level position as a data scientist? Or at least an interview? How do I get in front of prospective employers? I’m in the West of Scotland.

I'm trying to change career and relocate after teaching English abroad for 8 years. I think that data science is for me: I can program (C, Java, R, HTML5/CSS/JS, a couple of apps on Google Play), I'm pretty numerate (I have a degree in Physics), and I like getting to grips with data and communicating.

I'm most of the way through the Data Science specialisation track on Coursera, and so far I'm enjoying it and doing well and managing to answer questions and help other people on the course forums.

Things I’ve tried: going to a local tech meetup (for startuppers), lots of jobsites

Things I’m interested in: open data, linguistics, infosec, social inclusion

I’ve been self-employed in a foreign country for a few years, so references are hard to come by, but surely it counts for something.

Have you done something similar? Do you know what employers are searching for? Your help is appreciated!
posted by Wrinkled Stumpskin to Work & Money (6 answers total) 19 users marked this as a favorite
Best answer: I think what you want to look for is "data analyst".
posted by srboisvert at 4:57 PM on September 12, 2015 [1 favorite]

Best answer: A slow burner but what I did was find an open source project I was interested in in the field I wanted to work in and start contributing to documentation and bugfixes for that software. Actually job offers started coming in pretty quickly after I got serious with that but for various reasons I waited a few years before jumping into the field with both feet.
posted by singingfish at 8:19 PM on September 12, 2015

Best answer: If I were interviewing you, I'd want to be able to talk about projects you had done to get a sense of whether you have the data-reasoning skills that my data science team needs. In particular, it would be good if you had one or two independent projects that were bigger than a cousera class project - I like those for learning skills but I also think they tend to be a little too pre-baked to really show off your decision making.

I agree that data analyst roles can be a way to get into the industry, if you're having a hard time getting looked at for data science roles.
posted by anne_severson at 2:02 AM on September 13, 2015 [1 favorite]

Best answer: Hello from Edinburgh! I work at a tech company here and look after the Data Science team. So far we've hired two Data Scientists and here's some of the things we look for:

* professional/educational or a hobby experience with interrogating large data sets to answer questions - it's even better if the large data set is particularly messy and needs a lot of rationalisation before you can use it.

* a background in Physics would be very important to us, or some other strong background in statistics.

* programming experience and good examples of using any programming language to manage data and analyse data would be very helpful.

* commercial awareness - we're looking for people who want to ask and answer questions to solve real-world, commercial problems and would be asking questions like "what kinds of questions do you think we'd like to answer at this company?"

* strong communication skills - can you explain your analysis to a room full of people who have no understanding of statistics? Are you able to sit down with anyone in the company and find out what they'd like to measure, then help them put together a project to measure it?

For us we've come across more people with the physics background and even commercial awareness, but maybe without the strong programming background (years of experience). I'd be very interested in someone who could say "I have a strong understanding of the analysis methods and also here are some examples of how i've (for example) used log files to construct a data set and then conducted an analysis that led to X conclusion). As other people have suggested, maybe working on an open source project or some kind of personal project could help you get that kind of experience.

In terms of how to get a hiring manager's attention, I think a covering letter that shows an understanding of the kinds of questions that Company X may be trying to solve would go a long way.

We aren't hiring at the moment, but if you MeMail me i'd be happy to let you know when we are. Other than us, there are some great tech companies in Edinburgh (i only know Edinburgh, sorry!) who you could send a speculative CV to even if they aren't advertising for a position. Assuming you are interested in working on the East Coast, I would suggest following the people who run Codebase and going to their metopes as well.
posted by ukdanae at 2:59 AM on September 13, 2015 [3 favorites]

(that should be "meetups" not "metopes"!)
posted by ukdanae at 3:05 AM on September 13, 2015

Best answer: Seconding ukdanae. I've been hiring data scientists lately too and that mix of skills is quite like what we look for.

One thing I'll add - there is (in my mind) a growing split within data science between decision-making + data products. The decision-making branch is a more sophisticated and empowered version of business-intelligence type role, whose internal collaborators are often more product-focused. Data products, on the other hand, are rarer (fewer companies need people doing sophisticated machine learning, modeling, recommendation, clustering, etc) and much more technically demanding. Amazon's product recommendation team, for example, basically recruits only PhDs from particular research areas. People in this role need to be much more technically strong, but don't need to be the collaborators and communicators found in the decision-making side of things.

The decision-making branch is much more accessible to someone in your shoes. We interview pretty widely for that role and are not dogmatic about particular educational backgrounds provided you're a strong and quick quantitative thinker, have good product sense (you need to be able to ask appropriate research questions; this is much rarer than you might think), and be a strong communicator to non-quant audiences. Programming in any serious way is not important to us, but most companies do view that as the price of admission.

That said, you will probably need to find experience outside of places like Coursera. I don't know why, but courses like that are typically laser focused on prepping people for data product roles that people without a much more elaborate educational background are very unlikely to get. If the coursework involves KNN, random forest, neural networks, "deep" anything - it's probably irrelevant for decision-making work. It's too time consuming and hard to explain to a lay audience. If we do it, it's strictly for vanity purposes.
posted by heresiarch at 10:51 AM on September 13, 2015 [1 favorite]

« Older I Broke Up with Him and I'm Heartbroken   |   Stuck in the middle with you Newer »
This thread is closed to new comments.