As a major component of the foundational HCI course in user experience design methods, myself and a team of four other MHCI students were privileged to have the opportunity to collaborate with Intel Labs Pittsburgh. The goal of the project was to understand the workflows of administrators and users of the Hadoop cloud computing platform to aid in developing tools for determining the cause of errors and poor performance. Our team conducted user interviews, synthesized key findings, built click-through prototypes of a new unified Hadoop interface and finally conducted usability evaluations through further interviews and keyboard-level modeling on the resulting prototype.
Our research lead to insights that divined a better understanding of how users work in cloud computing environments and how they approach problems that they encounter. Our proposed solution integrated aspects of filesystem management, command-line functionality, web-based task monitoring, and overall system health indicators. As a team member, I participated in the entirety of the user research, synthesis, prototyping and testing stages. I was primarily responsible for creating keyboard-level modeling simulations, which turned into a testable prototype on which think-aloud usability tests were performed based on tasks that I defined.
The findings from this project were assembled into a paper, titled Understanding and improving the Diagnostic Workflow of MapReduce Users, that was accepted into the 2011 ACM CHIMIT conference.