News

How A New Review Process Won the Wharton Hack-AI-thon, Presented by Expedia Group

Now in its third year, the Wharton Hack-AI-thon, presented by Expedia Group, brought together some of Penn’s most ambitious builders to tackle real-world challenges at the intersection of artificial intelligence and business. This year’s competition was no exception, marked by rapid prototyping, late-night debates, and a final round full of energy, precision, and creativity.

Among a highly competitive field, one team stood out for their technical rigor, clear product thinking, and ability to execute under pressure.

From Classroom Collaboration to Hack-AI-thon Champions

The winning team, WayWise – Utkarsh Lal, SEAS’27, Vamsi Krishna, SEAS’27, Vedant Zope, SEAS’26, and Prithvi Seshadri, SEAS’27 – didn’t meet by chance. All four are graduate students in Penn Engineering, spanning computer science and data science programs, and had already built a strong foundation working together.

“We’re all friends at Penn, in similar programs and classes, and we’ve collaborated in hackathons before,” explained Lal. “That prior experience meant we already had strong chemistry, which helped us hit the ground running.”

This year’s business challenge from Expedia Group tasked teams with creating a tool to enhance the review process both for existing guests leaving reviews on properties and for future guests seeking current, complete information before they travel. WayWise approached the problem from a systems-level perspective. Rather than building a flashy prototype, they focused on something more ambitious: designing a solution that could realistically operate at scale.

Their product reimagines how platforms collect and refine user reviews. “We focused on two key requirements: efficient inference and deterministic ranking,” said Zope. The team developed a dynamic questioning pipeline that identifies gaps in existing reviews and asks users targeted follow-up questions. Their system evaluates:

How much information already exists
How recent and reliable that information is
Whether conflicting signals are present

By prioritizing these gaps, their system maximizes information gain, ultimately producing more accurate, consistent, and up-to-date reviews. It’s a deceptively simple idea with significant implications: better data collection leads to better recommendations, which is core to platforms like Expedia.

A few people seated in a conference room, engaged in listening and taking notes. — Georges Haddad from Expedia Group watches final presentations

Two Days, Countless Decisions

Like many Hack-AI-thon teams, WayWise’s journey was defined by effort and commitment.

“It was pretty intense. We built everything in about two days,” said Seshadri.

Those two days included late nights, rapid iteration, and even unexpected debates, like whether to use AI to generate their slide deck, a discussion that “probably took longer than it should have,” but became one of the more memorable moments of the experience.

Technically, one of their biggest hurdles was ensuring consistent multilingual sentiment analysis. Integrating a DeBERTa-based model into their backend required careful tuning to maintain reliability across languages – no small feat under time pressure.

At the same time, the team faced a classic hackathon challenge: ambition versus time.

“There were a lot of things we wanted to build, but we had to stay focused on what actually mattered for the demo,” said Seshadri.

The Final Pitch

By the time finals arrived, the team was running on a mix of exhaustion and adrenaline.

“After two days of hacking, we were definitely tired,” said Lal.

Once the team was notified that they were finalists for the competition, and that they would be presenting their solution to a panel of judges and Expedia Group representatives, they shifted gears. The group tightened their demo and refined their narrative, which in turn helped to shore up their confidence heading into the final day. “We felt we had a solid shot, so we focused on delivering a strong pitch.”

That focus paid off.

A Winning Solution

After the third and second-place teams were announced, WayWise learned they had come out on top, and would be receiving Apple products, OneKeyCash (the rewards currency at Expedia Group), and a chance to present in front of the Wharton AI & Analytics Initiative’s Advisory Board.

“Utkarsh immediately jumped up and yelled ‘YES,’” the team recalled. “Then someone said, ‘We get Apple Watches,’ which got everyone laughing.”

What Comes Next

While the Hack-AI-thon is a competition, it’s also a launchpad. For this team, the journey may not end here. “The team at Expedia Group mentioned they may follow up,” Krishna said. “We think there’s real potential to build on this.”

Their solution’s lightweight, scalable design makes it a strong candidate for real-world application, exactly the kind of outcome the Hack-AI-thon is designed to inspire.