As part of my job at Devoin I worked for Borusan Oto to recreate the Caterpillar's database from heavily JS-based online catalogue. I used ingenious scraping/reverse engineering techniques to rebuild this million-data points database.
Customer Service department of BSH had two reporting systems: SMA and QPR. Data from both of them was exported into Excel and manually processed. I built an automation system using VBA (Visual Basic) and made it possible to finish monthly reports in a few minutes. Press here for a glimpse to non-NDA contribution.
A frozen project. It aims at creating a universal course equivalence database among all the undergraduate colleges in the world. The early prototyping works but the project has continuously been postponed in favor of other more urgent ones. The collaboration requests are welcome.
Stata community has been lacking a panel fully-modified ordinary least squares regression command. I have built it and it is now available as an SSC package here. The package has found its demand and is being heavily used.
I developed an algorithm and corresponding R function that uniformly generates correlation matrices for any n-factor models in factor analysis. This contributed to a psychometric research by a client.
Most of the crypto research has been going in blogs instead of academic journals. This caused most of the incentive structures to be literally verbal. I created a mathematical model of token economies and a prototype R simulation to make future analyses rigorous.
I have successfully tutored a number of graduate economics students in their coursework and their theses.
I proposed to use Kamada and Kojima's matching mechanism to efficiently allocate Hajj visas to worldwide pilgrims. The process will significantly improve the current mechanism and completely solve the problem of unfilled quotas. For details press here