Into data before it was big. Keeping the world safe from folks armed with pivot tables and people trained to detect cats at scale. A great man for things that do not yet exist. Contact me if you are working on interesting problems and want to talk to an economist. Shallow learning for the win.
Please do not e-mail me with questions unless we drank at least one beer together. I am happy to help, but I won't do it over e-mail, LinkedIn, Facebook, Twitter, etc., as I much prefer to have my answer be a public good, accessible to everyone in the future. Moreover, if I am wrong, someone will usually come along to correct me, which is much better for everyone involved. Create a simpler reproducible example on one of the toy data sets that demonstrates the problem you are having, put that up on SO, and then tag me. I will take a look when I get the chance.