NYC area developers gathered for a hackathon in SoHo on December 6th to build with AssemblyAI’s industry-leading Speech AI models. This in-person event was hosted by our Developer Relations team to bring together coders of all levels. Regardless of attendees experience with our tools, free credits and on-site support helped every attendee hack away with creative usage of our products.
What's a hackathon?
Peter McKee, head of Developer Relations at AssemblyAI, kicked us off with our main goals for the day:
1. Create something using AssemblyAI’s leading Speech AI API
2. Connect with developers and AI enthusiasts in the community
3. Healthy competition, food, coffee, and swag
Time for creation and innovation
From there, we jumped right into hacking. Teams, many of whom met and formed spontaneously, spread out across our offices and got to building. While some teams came with pre-formed concepts, we heard live pivoting after they better understood the full scope of our API's offerings.
The AssemblyAI team helped attendees navigate through our Getting Started docs and answered questions:
"How can I build out a project with interview-style conversations?"
"We'd recommend you get familiar with our Speaker Diarization model!"
"Do I need a Go-to-Market plan?"
"Not required, but certainly welcome!"
"I'm thinking of building a chatbot for a customer support tool, which product should I use?"
"You might want to start with Speech Understanding to leverage LLM capabilities!"
"What's better for our presentation, a live demo or a recording?"
"We don't let perfect be the enemy of cool. We'll always prefer a live demo!"
We saw teams grab whiteboards, portable monitors, and scratch pads to map out their ideas. One attendee even brought a microphone to capture their own voice for their work with our speech-to-text models. You could practically see the creative juices flowing! You could also actually see the coffee cups, snacks, and energy drinks on the table that energized everyone.
Around 1pm, pizza and swag rolled in. Hackers took a quick break to re-fuel and grab an AssemblyAI t-shirt. During this time, we reminded everyone of project submissions that were due to be presented in a couple of hours. We wanted teams to focus on the work they completed, rather than overly detailed slides, so these submission requirements were kept to a minimum: tell us who you are, what you built, then show it off!
Show us what you made
Before we knew it, it was time for project presentations. I grabbed a microphone to emcee and AssemblyAI VP of Marketing Christy Roach, Senior ML Developer Advocate Smitha Kolan, and Peter McKee were in the front row for the judging panel. Teams presented the work they completed under the tight time constraint and took questions from the audience. We saw a huge and creative range of projects here from undergrads to very senior developers that successfully embodied AssemblyAI's vision to supply superhuman Speech AI models that unlock entirely new classes of applications and products to be built leveraging voice data. We love seeing all the different use cases of our products, especially in innovative events like this.
And the winners are…
The most difficult part of the day: awarding just a few teams from a large group of inventive builders.
The winning project: Dealty by Slavik Kaushan and Mario Uribe! Slavik and Mario are developers, entrepreneurs, and AI enthusiasts who used AssemblyAI’s streaming speech-to-text and entity detection models to build a Speech AI tool to facilitate real estate investment deals. First, they generated a real-estate deal call using OpenAI. AssemblyAI’s streaming was then used to transcribe the call and our entity detection endpoint retrieved field values (type of deal, property type, finances, etc). Finally, Dealty’s UI was populated with the field values in real-time. They are currently in stealth mode so be on the lookout for their next venture.
"We leveraged AssemblyAI to seamlessly process conversation recordings in real-time, enabling us to populate application forms on the fly. By sending audio to their API, we received fully processed and structured data in a single call—a highly convenient solution for any real-time processing needs," they explained.
The runner-up project was Muse by Arjun Sivakumar, Ritvij Saxena, and Siddharth Prabhakaran — a mental health journaling assistant that the AssemblyAI team agreed we’d buy in the app store immediately. More of their work can be found here: Arjun Sivakumar, Ritvij Saxena, and Siddharth Prabhakaran.
Finally, we gave a special shoutout to project Say What by Stuart Allen, Nikki Hu, Kirk Enbysk, and Peter Shrieve-Don. They built an AssemblyAI-powered guessing game that turns audio clips into transcriptions for interactive learning. You can view more of their work here: Stuart Allen, Nikki Hu, Kirk Enbysk, and Peter Shrieve-Don.
Connection doesn't stop here
We wrapped the day with team photos and a happy hour to celebrate. Further questions about project methodology were asked, LinkedIn info was exchanged, and accolades were given.
Hackathons are just one way we like to engage with the developer community. If you want to get further connected with us, we recommend you join our Discord and sign up for our newsletter. See you at the next one!
