Better Seattle biking? A UW student and a high schooler have it mapped

Jared Hwang knew just who to turn to for the complicated work of corralling artificial intelligence to help power his doctoral project at the University of Washington.

A high school student in Issaquah named John O'Meara.

Together they made Bike Butler - an internet mapping tool that caters specifically to Seattle bicyclists. The site allows people to tinker with way-finding criteria and customize what route works best for them - from a leisurely weekend family ride in protected bike lanes all the way to the quick dash to work next to traffic.

It was a very selfish motivation: I've been wanting to ride my bike more," said Hwang. "I know the benefits of biking - for health, for sustainability. But every time I wanted to bike, Google Maps would just be terrible."

Hwang's map comes at a right time, and not just because May is Bike Everywhere Month. Seattle is in a frenzy of bikeway building, connecting the city with a network of protected bike lanes, trails and bike-oriented Healthy Streets.

The build-out has helped Seattle become one of the best cities in the U.S. for people on bikes, with record numbers of riders and miles ridden tallied by the fitness and social media app Strava, Lime and its rentable bikes, the U.S. Census Bureau and the Fremont Bridge Bike Counter, which saw more than 1 million bikes cross it in 2025.

As more people take to two wheels, however, they quickly realize that not all Seattle streets are created equal - as Hwang did when he plotted a commute from his Capitol Hill home to campus.

All of the common map apps - Google, Apple, Strava, Ride with GPS - told him to ride north, up Broadway toward the University Bridge.

"On paper, Broadway sounds great. It does have a bike lane," Hwang said, noting that he took a lot of inspiration from Google and the other routing tools. "And a bike lane is better than other routes."

Still, Hwang knew there were better options: quieter, shadier, safer streets with smoother pavement. So he got to work.

He wanted to find a route based on his preferences, knowing that they may change depending on time of day, day of the week and even season. In the summer, he may want to take the long way home from campus, following shady streets. On a rainy day in autumn, he'll probably want to get home as quickly as possible where any reasonable road will do.

To zero in on individual bicyclists' likes, Hwang built a map allowing users to choose defined route profiles like "slow and safe" and "fast and loose."

But he wanted more personalization, so he included eight modifiable bikeability factors on a sliding gradient, including bike lanes, elevation change, greenery and pavement quality.

With this information, Bike Butler generates a route. It identifies segments along the way where the ride changes character - for example, if the route turns off a road with a protected bike lane to one with no bike lane.

Images of the street show the rider what awaits, and includes notes describing how each segment aligns with their preferences.

"There are so many different reasons why we might be biking," said Jon Froehlich, Hwang's adviser at UW's Paul G. Allen School of Computer Science and Engineering. "Computers treat all those reasons as the same."

Bike Butler is different, he said. It recognizes the human variety on wheels.

"I had kids, and that suddenly changes your perspective on traffic and riding a bike," he said. "Safety is paramount. Maybe I'm not going to take the same route here."

Human-centered AI

Hwang knew some of the things that make a city bikeable and why. Cyclists generally prefer routes with bicycle infrastructure - bike lanes, trails and greenways - slow cars, low traffic and not a lot of hills.

With those factors in mind, Hwang first turned to publicly available government datasets and the community-produced OpenStreetMap.

But his prototype didn't speak to more subjective attributes like quality of pavement, level of greenery and overall "visual aesthetic of a street scene," as he wrote in a research paper about his project. For that, he turned to O'Meara.

"I poached him," said Hwang.

O'Meara, who will be a UW freshman in the fall, was already at work in Froehlich's Makeability Lab at the university.

O'Meara had scored a coveted role at the lab by contributing to its Project Sidewalk, which aims to improve how cities collect and use data on pedestrian architecture, like sidewalks, to help support equitable urban planning and expand access for people with disabilities.

Froehlich said O'Meara was just one of the many high schools student who send him requests to work in his lab each year. "But he distinguished himself by the actual work he was doing" on the open source Project Sidewalk, Froehlich said.

For that, O'Meara analyzed images using artificial intelligence to find and rate curb ramps.

Hwang, who also works in Froehlich's lab as a second-year doctoral student, met O'Meara there, and knew the teenager had AI skills he didn't. So, Hwang recruited O'Meara to do something that seems a bit contradictory. He had O'Meara train Google's Gemini AI to think like a bicycling human as it processed and analyzed 55,000 Google Street View images.

Hwang knew AI could "scare people off," but he said it was necessary to the project and was used in constrained way.

Now that the map has been tested and improved by a group of cycling volunteers, any Seattleite can generate a personalized bike route.

The map isn't perfect, as Hwang noted. As a research project, it won't be regularly updated, so it's "frozen in time" and won't reflect when the city adds a new bikeway. It also has no "intersection awareness," leaving riders on their own where two streets meet, a notoriously dangerous part of any bike ride. And it's hard to read on a smartphone.

Still, it gives bicyclists a better way. Hwang's commute on Bike Butler, for instance, avoids Broadway and instead takes him on quieter neighborhood streets and over the Montlake Bridge.

By offering safer routes - or just highly personalized ones - Hwang hopes his map does more than get people from Point A to Point B.

"I want more people to bike," he said. "The more people bike, the better it is for everyone."

But, personally, he has bigger ambitions for his map.

The map's underlying code is freely available online, so anyone can build a Bike Butler map for their city, if similar data is available there.

Locally, Hwang hopes to keep Seattle on its bike-friendly trajectory. If enough people use the map and tinker with their preferences - showing what streets they use or tend to avoid, and why - Hwang said he could pass the data to the city, which it could use to better understand cyclist behavior and better plan the city's bikeways.

"If this ends up making every street (get) a bike lane, I'm all for it, Hwang said.

Copyright 2026 Tribune Content Agency. All Rights Reserved.

This story was originally published May 12, 2026 at 6:37 AM.