Monday, November 4, 2013

The Center for Urban Science and Progress

The below post was submitted by an anonymous guest contributor:


I'd like to share with this community what I learned from a talk about the recently-established Center for Urban Science and Progress at NYU. Steve Koonin, the Center's director, is a MIT-trained physicist with a strong track record in both research and public policy, serving on the Caltech faculty and as Undersecretary for Science in the Department of Energy, among many other things. Steve gave an engaging talk presenting examples of CUSP research as well as a sales pitch for his new center.  If you are interested in non-astronomy-professor job opportunities for astronomers, read on.

Aside from the fact that most of us live in cities and thus care at least a little bit about urban planning and infrastructure, I found his presentation to be of particular interest from the standpoint of applying the tools of astronomy to problems cities face.  There are currently enormous -- and growing -- data sets characterizing the urban landscape, ranging from images of various parts of the city to GPS tags on taxis and cell phones to numbers collected by public utilities.

Image analysis techniques from time-domain astronomy can reveal usage patterns in images of city lights at night.  Understanding the properties of celestial objects from multi-band photometry is similar to extracting information about a landscape from hyperspectral images. Steve showed one example of a thermal infrared image in which you can clearly see different temperatures in different apartments.  With a similar image in several passbands, you could probably determine everyone's thermostat settings!  Large sky surveys and space missions have already addressed many of the issues that appear in organizing and archiving large data sets, data products, and associated metadata.  Any student working on a large survey is skilled in semi-automated analysis of these large data sets.  Further, astronomers are very good at defining meaningful scientific questions that can be answered through observation and making inferences about scientific questions from quantitative observational data.

One of CUSP's strengths is that its leadership recognizes the multidisciplinary nature of its research questions, and is currently recruiting a diverse team that includes astronomers, physicists, machine learning experts, data scientists, social scientists, urban planners, … from academia, industry, and government.  I see this as a real opportunity for astronomers who are interested in using their technical skills to address meaningful research questions outside of basic astronomy research.  It is applied rather than basic research, it is for the public good, and has a timeline for translating research results into demonstrable impact on society measured in years rather than decades.

The CUSP website is the first place to look for job opportunities, but Steve said to email if you're interested and there isn't a position listed (postdocs, this means you).  And, of course, CUSP isn't the only place where this type of work is done.  More broadly, if you are interested in this type of opportunity and you are:

  • An undergraduate student: Take a look at CUSP's master's program.  PhDs are great, but they aren't necessarily the right path for everyone, nor are they necessary to have a successful career in many fields.
  • A graduate student: Make sure you have a solid skill set in addition to a solid understanding of your science.  Have a deep understanding of any statistical analysis you do, take the time to write clean code rather than code that is a mess but gets the job done, understand the principles behind the image analysis you use even if a pipeline does most of the work for you.
  • A postdoc: Be able to articulate the value of your skill set as well as you can articulate the value of your scientific work.  Have an elevator speech that explains what you can do and why it's useful.