Social Science PhD studen to Data Scientist?
June 17, 2017 1:11 PM   Subscribe

I'm looking for information about the path someone in my position would take to work as a Data Scientist. Salary? Quality of life? Length of projects? Boot camp? Etc.

Currently, I'm a PhD student in the social sciences at a top Ivy school. We always get sent around fellowship opportunities for these data science-for-PhD-students-looking-to-work-in-industry-bootcamps, which are funded. I have some questions about what this path would look like for someone in my position.

I have good quant skills, but not great. I know R well, have taken advanced stats courses, but I don't do anything like derive new estimators or improve an algorithm in my research. I've taken a couple Machine Learning courses as well. I understand these concepts, can implement these tools, but am far and away from ever developing my own.

A lot of these bootcamps include testimonials from physics, engineering, and math PhDs. I'm not one of them, and simply do not have their technical ability. Although the bootcamps say the requirements are low and they will train you on the most recent tools that data scientists use.

What's the life of a social scientist turned data scientist look like? Giving up the freedom that academia provides, and the ability to work on a long term project, would be hard. Do data scientists generally work on longer term things? I know at the big name companies they do and are given freedom. Are those benefits limited to the top?

What's the salary as a data scientist look like? These bootcamps say the vast majority start with around 100k salary. How fast does the salary increase? does it level off? If I became a data scientist, does that usually mean I would stay in a similar position for the foreseeable future? or do people move up in the company?

I fully intend to go into academia, but as the job market grows nearer I want to explore my outside options. Thanks!
posted by MisantropicPainforest to Work & Money (7 answers total) 9 users marked this as a favorite
Best answer: I'm a CS PhD working in a mid-sized company as a machine learning scientist, and I work with data scientists with your background. You absolutely do not need the quantitative chops of physics and math PhDs... it actually surprises me that they tend to do data science, because it's more about understanding the domains and making models -- along with a good stats grounding, of course -- rather than mathematical heavy-lifting. You'll be absolutely fine.

As far as freedom, it varies, but where I am, different teams hire data scientists. Once you're in the company, you can choose which team you want to work with (assuming there are openings), and have quite a bit of freedom within the project, though you do have to collaborate more closely with your colleagues compared to academia.

I suggest you apply to places where you can work on projects that are related to your field of research (if possible), or otherwise interest you. Your domain knowledge will be useful, and you'll presumably enjoy it more than a problem that's totally unrelated to what you've been studying for the past N years.

I'm biased, but "younger" industries like tech have working styles more similar to academia -- casual environments, liberal-leaning colleagues, flexible work-from-home policies. Most importantly, they're generally up to date on statistical techniques and programming tools. Traditional places like banks or older companies have their advantages, but if you're not a fan of dressing formally, clocking in your hours, having to explain the difference between linear and logistic regression a hundred times, or being prohibited from using R/Python because they're welded to SAS/Stata, you might find the transition hard.
posted by redlines at 1:53 PM on June 17, 2017 [2 favorites]

As for mobility, I think you can move up to more supervisory positions (manager, VP of engineering, etc), but it's hard to say what the exact trajectories are because "data science" hasn't been around that long, and we don't know what the future holds.
posted by redlines at 1:58 PM on June 17, 2017

Best answer: I have a career that falls into the data science bucket, at an established non-tech company in the midwest. I have a pure math background and have really never applied any of the math I learned in school beyond stats/probability.

I think the biggest thing to understand is that data science is a very, very broad term, that covers many different topics, from data visualization to optimization to machine learning. Data scientists are employed by all sorts of companies, large and small, tech and non-tech, ones with disciplined approach to data management and ones which do no. It's also not just about number crunching. Good data scientists will have a deep knowledge of whatever domain they're working in, and will be able to clearly communicate technical concepts to their non-technical customers (whether internal or external). I imagine a social science PhD would be good on both those fronts. There are definitely organizations out there doing data science work in the social science arena. They may not be the next "cool" tech company utilizing highly advanced machine learning algorithms, but they exist, and they exist all over the country too. If you know R and have stats knowledge, I think you're probably good without a bootcamp, assuming you're flexible on the type of data science job you're looking for.

I can't speak to academia vs corporate life (I jumped ship on my PhD program and exited with a masters pretty quickly when I realized academia, in general, wasn't for me), but I have a pretty decent work/life balance, and have excellent benefits. I started my career about 7 years ago and have moved into management in that time. There are some longer term projects, but I usually have multiple small-to-medium sized projects going on at the same time. Stress levels can get high at times when there are a lot of fires to put out on very short notice, but in general my work stays at work when I leave for the day.

I think whether there are opportunities for advancement and growth really depend on the type of company. For instance, the analytics team at my company is just a relatively small department inside a very large operationally focused organization. This provides opportunities for senior level analysts to branch out into other departments (usually in management) if they want to. At a company which is solely focused on analytics, there might be fewer opportunities for advancement, but more potential for lateral moves to different teams.

As for salary, I got significant bumps with each of my promotions and we also get annual bonuses, but annual base increases are pretty small. I think you'll find that at most companies. The biggest salary gains usually happen when you change jobs. Also, do take into account cost of living when you're looking at salaries. From the research I've done, I've found that the salaries in my midwest city are better than the big coastal cities after adjusting for cost of living.
posted by noneuclidean at 3:06 PM on June 17, 2017

Best answer: I have a math PhD and my job title is "data scientist". For better or worse, it's a pretty meaningless job title--my job is probably much more akin to redlines's than to the people at their company whose title is "data scientist". I did not go the bootcamp route, though I know people who did and it seems to have worked out well for them. I've ended up far towards the engineering side of the data science spectrum (a term I just coined), whereas the the people I've known who went to bootcamps have ended up more towards the statistics/analysis side or somewhere in the middle. I don't know if this is an artifact of our interests/skills/personalities or an artifact of bootcamps vs choose-your-own-adventure.

Realistically, you have a much stronger statistics background than I did. I took a single stats course the year I finished because I realised I didn't want to go on the academic job market and figured it would give me some credibility. I'd written a fair bit of Python code for my research and had contributed a bit to open source math software (which sounds a lot more impressive than it actually was). There's generally little to no development of new techniques--it's mostly about figuring out how to apply something fairly well-known to the problem at hand.

Length of project varies. At my first job, when I left, I was still working on a project that was explained to me my first day. There had been stretches of weeks or even months where I'd been so far ahead of the engineering work for the other moving parts that I didn't touch it at all and times where it was all I worked on. Or someone could say "Hey, we'd make loads of money if we could predict abc" and you spend a week and come back and say "We can't actually do this because of x, y and z."

What's the salary as a data scientist look like? These bootcamps say the vast majority start with around 100k salary. How fast does the salary increase? does it level off? If I became a data scientist, does that usually mean I would stay in a similar position for the foreseeable future? or do people move up in the company?

$100k sounds right-ish, though it depends on where you are. It's a little hard for me to know, as my starting salary was probably substantially below market. (I will give you numbers if you me-mail me.) I've found the career/promotion path to be a little murky, to be honest, because I've always been on data science teams that are "not quite software developers" and no one is quite sure how apply that progression to us. (For example, one way you might promote a developer is to have them move over to lead another team. But, at a smaller company, there's only one data science team (frequently with high levels of responsibility compared to developers of similar levels of experience) and it makes no sense to have me move to another team, even though I could.)
posted by hoyland at 3:09 PM on June 17, 2017

All of the hosts of Partially Derivative, a data science podcast, are data scientists. None of them came from a heavy math/engineering background. Chris Albon actually started off as a social science PhD like you (quantitative poli sci, to be exact), while the other hosts spend an entire episode discussing how they got into data science from their non-math backgrounds.

You might to give that episode a listen and I can't imagine it would hurt to tweet/email them if you have follow-up questions.
posted by Ndwright at 4:36 PM on June 17, 2017

I know lots of Stanford social science PhDs who work in data science at Airbnb/Apple/Facebook/etc. You shouldn't need a boot camp - if you have decent academic stats skills that should be plenty. Most people aren't building new algorithms but are doing standard analyses on relevant data for the company. Salaries typically start above $100k and can move up pretty quickly, particularly if you start running a team. You may also get stock options which can be worth 100s of thousands of dollars. But salaries outside of the Bay Area are probably lower and you have to contend with the very high cost of living here, which means the salaries don't go so far.
posted by pombe at 4:59 PM on June 17, 2017

The Insight Fellows program might be worth looking into. I know a couple of people who have successfully moved from academia via Insight.
posted by avocado_of_merriment at 5:34 PM on June 17, 2017

« Older Best way of cooling a house using window A/C units...   |   how do bipolar people avoid bad financial... Newer »
This thread is closed to new comments.