Armin Samii

RCV is a better way to run elections, but the results can be hard to understand, which is why good visualizations are crucial. RCVis has become the de-facto standard in all USA RCV elections and has been used to inform the visualizations used by NYT, WaPo, and Financial Times. The link below shows FairVote's endorsement of RCVis.

An app that brings GoPro-quality technology onto the smartphone, allowing bicyclists to record their bike ride in case of a crash. Through a partnership with the City of Pittsburgh, app users can also report any hazards they encounter on their ride to the City of Pittsburgh.

In collaboration with the City of Pittsburgh, as part of Dashcam for your Bike, we have created a map of hazards encountered by bicyclists using the Dashcam App while riding a bike in Pittsburgh. The map is available to the public and city officials, who use it to determine where to improve city infrastructure.

UberCheats is a way to detect if Uber drivers and couriers were underpaid due to a bug in Uber's distance calculation algorithm. This project was featured in Wired, Wall Street Journal, and Vice. We focused on privacy, and the data was never uploaded to servers. It used a Chrome Extension that analyzed your Uber statements, parsed out how much Uber paid you and for what distance it paid you, then compared that data to Google Maps' Distance API to determine if they miscalculated the distance you traveled.

UberCheats can automatically parse thousands of trips at once, so a user could open their weekly statement and, within seconds, see a list of all trips they were underpaid for. They could then call Uber support, cite each list, and have all the data necessary to get paid correctly.

I was the lead author of this cloth simulation library used by thousands worldwide and the author of the corresponding academic paper published at SIGGRAPH Asia. I gave the talk in Singapore, introducing the paper to several hundred of my colleagues in computer graphics.

ARCSim uses finite element simulation with an adaptive triangle mesh: it predicts where the cloth needs more resolution based on several properties, including proximity to a collision, compression, and stretch, and adds that resolution before it is required.