Vaasavi Unnava began working as a Resident Assistant (RA) in the Mudge House on campus during her sophomore year, working one-on-one with students to ensure they have the best transition into the college lifestyle, from putting on educational programming to acting as an emergency responder.
It wasn’t until her second semester as an RA that Unnava, BS ’17, realized how inefficient the RA matching system was.
“It was in Professor (M. Bumin) Yenmez’s Market Design class that I began to notice trends within the RA matching system that weren’t very efficient,” Unnava said. “It was really a problem with the way the auction system was set up.”
Among the problems that Unnava noticed: redundancy, mismatches with staff, time loss – such as housefellows beginning placement conversations with each other long before the central conversation commences – and difficulty determining which housing communities were willing to commit their resources to particular RA candidates.
It was during Yenmez’s lesson on market unraveling that Unnava, an Undergraduate Economics student, began to think of ways to optimize the RA matching process. Unraveling is a phenomena in which people don’t trust a matching system. That semester, she wrote her final paper on the topic, proposing a new process.
“RA selection is a very challenging process in our office because it involves a large number of staff members and candidates. As a Community Advisor and former Resident Assistant, Vaasavi was uniquely situated to take what she was learning at the Tepper School and apply it to this problem,” said Bryan Koval, Coordinator of Student Life at Carnegie Mellon.
With the new system, which Unnava describes as a “simultaneous ascending action,” all RA candidates would be considered simultaneously. Each housefellow would receive a number of points directly proportional to the number of RAs that they need for their building, and they can use the points to bid on candidates as they wish.
For example, if a housefellow has 50 points, and they really want one particular RA on their staff, they can bid all 50 points on that person. However, in this case, they now have no points left to bid and will be given whichever candidates are remaining at the end of the process. Other housefellows may choose to spread their points across several candidates, with the possibility of being outbid. During each round, a housefellow can only bid on as many candidates as they have open positions.
“Many efficient matching algorithms do not allow for diversity as a large component of the hiring process, but in this process the style of the team built is entirely up to the housefellow,” Unnava explained in her proposal. “This is something that occurs in the original process as well, so that every candidate gets equal consideration, but would move considerably faster in the new process.”
If it turns out that multiple housefellows are equally committed to the same candidate, then the number of points that they would give up to secure this candidate will increase by one. After that, each housefellow will have one to two minutes to reconsider the set of individuals to whom they would like to commit. Once the time expires, the next round of commitments begins.
“One of the great things about this system is that housefellows can change which candidates they choose to commit to at various points in the process,” Unnava said. “By allowing housefellows to change the style of commitments, they continue to have a diverse staff while still being able to track who they currently have committed to having on their staff.”
The process ends when each candidate has only one individual interested in committing to them. At this point, the complete staff of a house has been chosen for every community.
This semester, the housefellows ran a pilot of Unnava’s program across campus, and hope to put it into practice next year.
“The pilot was a success… where the matching process used to take a full day, using this system it only took about one hour,” Unnava said.
“We appreciated the initiative and commitment that Vaasavi demonstrated by creating an RA matching system and piloting the system with the housefellows,” Koval said. “We are eager to take what we have learned from Vaasavi and this process and using it to further improve the RA selection process.”