It started on a quiet night at Monument Circle. Our crew was restless, leaning against our bikes under the glow of the fountains, half-joking about riding all the way to Indianapolis Motor Speedway. The problem? Too far, too late, and one ride wouldn't cut it before the free window ran out. Then someone said it: "What if we just built the app?" That was it. That was Relay. What started as boredom turned into a full-on sprint. Not to the Speedway, but to rethink how people actually move through cities. Not just fast. Not just convenient. Actually free. Relay is a real-time bikeshare routing engine that breaks one long trip into a series of optimized short hops, each one staying inside the free ride window. Instead of asking "what's the fastest route?", Relay asks something better: what's the smartest route given how the system is actually priced? We stream live station data through the GBFS feed, model the entire bikeshare network as a weighted graph, and run a constraint-based shortest path algorithm that keeps every ride under roughly 7.5 km. The result is a seamless multi-hop journey: walk, bike, dock, walk, bike, dock, destination. Mapbox powers the turn-by-turn directions so the whole thing feels effortless, even though what's happening under the hood is anything but. Building it was genuinely hard. Station availability flips constantly. A perfect route at 9:01 PM can be broken by 9:02 if a dock fills or empties. We had to design a system that adapts in real time, not just computes a static path and calls it done. We were constantly wrestling with tradeoffs: walking distance versus ride efficiency, availability versus optimal routing, accuracy versus real-time performance. And beyond the algorithm, there was the usability problem. A technically perfect route that confuses people is worthless. We obsessed over making multi-hop navigation feel natural and worth following, because the best solution is always the one people actually use. We're proud that we turned a late-night spark into a working, real-time routing system over a single weekend. But what we're really proud of is that it changes the math on urban transit in a meaningful way, making bikeshare actually free for trips most people would write off as too far. The biggest thing we took away is that optimization isn't just about algorithms. It's about understanding how real-world systems actually behave, and building something that moves with them. Small inefficiencies in everyday infrastructure compound fast, and solving the right one can unlock real impact. We built Relay for Indianapolis, but GBFS is a global standard. Every compatible bikeshare city is a city where this works. The goal from here is simple: make genuinely free urban mobility the default, not the exception. Relay isn't just a routing tool. It's a smarter way to move through cities.

Share this project:

Updates