Learning git is not enough: becoming a data scientist after a science PhD
Learning git is not enough: becoming a data scientist after a science PhDHere is useful advice about moving from academia into data science after completing a PhD in a natural science.
How finding a job is different than in academia
The visa situation is much tougher
In most countries, universities find it relatively easy to get visas for students, postdocs and faculty. Tech companies generally find it much harder.
In the United States, H-1B “application season” is January through February. You need to have the offer by March so the employer can submit by the first week of April. You then have a roughly one in four chance of being selected in the lottery (or slightly higher if you did your PhD in the United States). If you’re selected, you can’t actually start work until October 1.
If you have very strong academic credentials, can afford to pay the filing fees yourself, and are patient then you may be in the running for an O-1 visa. This has upsides and downsides, but the upside is there is no lottery and no deadline. If you’re going this route, you’ll probably need your own attorney.
Self-promotion is still valuable
Compared to academia, it’s less necessary to give talks, write publicly and network. But it doesn’t stop being incredibly effective. So if it’s something you enjoy, don’t stop.
Changing jobs is easier and more common
A successful academic career consists of a PhD, one or two postdocs, and a tenure track position you remain in for the rest of your working life. Opportunities to correct course come up once per year, and don’t look good on a CV.
In tech, it’s normal to change jobs every couple of years. By all means try to get career decisions right, but they are far, far less fraught and irrevocable than in academia.
Not everything has an application form
I was an astronomer. When I was applying for jobs in academia I would go to the AAS Job Register, a website that lists every single vacancy in international astronomy. It’s a finite set. The employers have reputations going back centuries and I’d been reading my potential managers’ work for years.
When I left academia, that was no longer true. Like me, you’ll have to do huge amounts of research to find the vacancies and learn about the employers, their products and their teams. And you’ll never find all the vacancies, because many of them are never advertised.
I think Cathy O’Neil sums up the danger here well. She’s talking about Harvard grads, but this is also true of science PhDs:
[Harvard grads] are vulnerable to Wall Street investment firms and to things like Teach for America because they have application processes at all. But life, normal adult life, doesn’t have an application process.
When you leave academia you enter “normal adult life”. The lack of an application form is exciting and overwhelming. You’ll get the hang of it.
But until you do, one of the dangers is that boot camps will be seductive because they’ll feel like academia. In some cases they can be exactly what you need to jump-start your career. But they are not always necessary. If you’re planning to do a boot camp, think carefully and honestly about your reasons.
How to choose where to work
If you’ll be working in the engineering division, then make sure you’ll be working alongside experienced software engineers and data engineers.
And where ever you’ll be working, make sure they have data. There are times to take a job where this isn’t the case, but your first job after a science PhD is not one of them.
Think about how data science relates to the mission of the company. To that end, I love sharing this quote from Stitch Fix:
A Data Scientist should look for a company that actually uses data science to set themselves apart from the competition. When this happens, the company becomes supportive to data science instead of the other way around. It’s willing to invest in acquiring the top talent, building the necessary infrastructure, pioneering the latest algorithmic and computational techniques, and building incredible engineering products to manifest the data science. “Good enough” is not a phrase that is uttered in the context of a strategic differentiator.
I don’t agree that a data scientist “should” work for a company where data science is a strategic differentiator. Plenty of people have fulfilling careers and do great work when that isn’t the case.
But the distinction I think they’re getting at — do you want your work to be the center of attention or not? — is a very useful one to think about.
Finally, think about the impact you want to have on the world. One of the reasons I left academia was to have more of an impact. But the corollary of having more impact is having more scope to do harm. Data scientists enable surveillance culture and recapitulate discrimination at scale. For more on this, read Weapons of Math Destructionand Big Data’s Disparate Impact.
Original. Reposted with permission.
Bio: Mike Lee Williams is research engineer at Fast Forward Labs.
Related:
