Apartment hunters, accessibility tools and debugging agents top a 12-hour build day in Manhattan
By 9 a.m. Sunday, the room at 9th Avenue and 31st Street was already buzzing with excitement and anticipation. Laptops were open, coffee was poured, and the line for scallion pancake bacon egg and cheese wraps (catered by Win Son Bakery) snaked across the room. Twelve hours later, teams made from 150+ builders had their projects for science-fair demos at the inaugural AI Engineer NYC Hackathon in an unofficial kickoff for AI Engineer New York conference week.
Major League Hacking (MLH) presented the event with AI Engineer as co-host. It was free for anyone to join and bring their team, their idea, or just themselves and hunger to learn. Attendance at the hackathon exceeded capacity in the main 9th floor space, so an overflow room on the 6th floor was set up to accommodate the extra demand.
The crowd was not the usual hackathon mix of students. Many attendees build software for a living as engineers, founders, researchers and product people, and they chose to spend a Sunday building some more. If you haven’t been to a hackathon before, or in a while, the choice to spend more time “working” can feel strange. But nothing can replace the way it feels to get your hands on a keyboard and work through a problem side by side with other builders. At the AI Engineer NYC hackathon, this really came through. Conversations moved from "how does this API work" to "here's how we handle this in production." Teams that had met at the morning team-formation session were trading deployment tips and opinions on agent frameworks within an hour.
Dumplings and Demos
The Nom Wah lunch spread, a nod to the Chinatown dim sum institution, did more for networking than any icebreaker. Builders carried plates of dumplings back to their tables and kept talking, and newly formed teams sat down together for the first time without a screen in front of them. Mentors circulated all afternoon, and in the evening the happy hour gave a final burst of energy before the demo deadline.
There was plenty of laughter, some groaning at stubborn bugs, and the occasional cheer when a demo worked on the first try. Many people who arrived alone left with teammates, collaborators and a few new friends.
A Very New York Winner
First place overall went to Boroly, which takes on one of the city's most familiar headaches: finding an apartment. Boroly compares listings side by side and adds neighborhood context like local grocers, restaurants, and laundromats. It scrapes this local data and uses Google's Gemini to summarize what it finds, so renters can see what living on a block is actually like, not just the square footage or the number of bedrooms. (source)
Second place went to Claude Code Ripple, a tool for developers (a “Mod,” a new class of agentic plugins) who have changed one line of code and then watched something unrelated break. Ripple traces how a change spreads through a codebase, including renames, before it ships.
Resurface took third. It turns forgotten bookmarks into animated reels and uses those saved links to build a profile of a user's interests that AI agents can draw on. (source)
Partner Prizes Spark Creative Builds
Partner challenges gave teams more ways to win. Public Palate won Best Use of Nimble by combining restaurant inspection records with reviews, and the runner-up, Nibble, finds the cheapest grocery prices in your area. Explorin won Best Use of Gemini API with experience-based personalization of travel recommendations. RetailScope AI took Best Use of Snowflake with a tool for finding commercial real estate deals. Zig Actions Runner won Best Use of Tiger Data for a lightweight, local GitHub Actions runner. rezolution.si won Best Use of Sentry with an agent that investigates bug reports and generates fixes. MyVision, which helps visually impaired users detect nearby obstacles, won the Sentry raffle.
Setting the Tone for the Week
The hackathon was planned as a lead-in to the conference, because many of the coding agents, data pipelines and context-aware assistants that builders engaged with on Sunday leverage the same workflows and tools that speakers will cover over the next few days. Attendees who missed Sunday can find these builders around the conference all week. They are the ones who look tired but happy, and many of them now have a new project and a new team.
Top comments (1)
tr.ee/dev-to