Boombox — Music Collaboration

The problem

Boombox set out to let musicians collaborate remotely: record audio and video takes, layer them over each other's tracks, and talk in real time while doing it. When I joined there was an idea and a pitch deck; the company needed a working product — first to convince investors, then to serve real musicians.

My approach

I was a core developer from prototype through production over three years.

• The prototype — I built the initial iOS prototype that the founders took into their seed-funding round: recording, playback over a backing track, and enough of the collaboration loop to make the vision tangible.
• Real-time core — I architected the audio/video recording pipeline with react-native-webrtc, with Socket.io carrying both chat and WebRTC signaling over a single stateful connection.
• Media at scale — takes are large files recorded on mobile networks, so I built the storage layer on AWS S3 around resumable multipart uploads and a local-first sync queue — a recording is safe on the device the moment it stops.
• Stability work — I migrated the backend integration from GraphQL to REST without service interruption, and added unit, integration, and e2e test suites that sharply cut post-deploy bugs.

Challenges

• Musicians hear latency — layered recording is a synchronization problem; takes are timestamped against a shared session clock at capture and aligned at playback, with a manual nudge control as the escape hatch for Bluetooth output latency.
• The iOS audio session — recording while monitoring while playing a backing track means competing AVAudioSession modes; I centralized session state in one native-side manager and treated interruptions (calls, Siri) as first-class events.

Outcome

The prototype secured the company's seed funding, and the production app grew to serve 10,000+ users on the App Store and Google Play.

Visit boombox.io →