Job Overview
Job description
- Salary:
- USD 180,000 - 250,000 per year
- Location:
- New York City, United States
- Work arrangement:
- On-site
Role Summary
About Sandbar
Sandbar is an interface company in New York City. We aim to augment individuals so we can each think, act, and move more freely. Our team has built SW, ML, and HW products across Meta, CTRL-labs, Google, Apple, Fitbit, Peloton, and Equinox.
Our first product, Stream, is a self extension—a private voice ring and conversational interface. Stream has been featured in WSJ, Bloomberg, & Wired, and begins shipping in Summer '26.
Join us in creating technology that extends human thinking.
Responsibilities
- Develop, evaluate, and deploy STT/TTS models spanning cloud models to resource-constrained on-device settings
- Optimize inference pipelines for latency, reliability, and concurrency
- Work closely with other ML/AI engineers, infrastructure engineers, designers, and cofounders on existing and future products
Requirements
- 4+ years machine learning experience
- Experience developing audio / voice models is required
- Experience shipping ML-based products is required
- Passion for human-computer interaction
FTE Benefits
- Health, vision, and dental benefits
- Company-sponsored 401(k)
- Unlimited PTO and sick time
- Early stage equity
About the Company
We’re looking for a machine learning engineer to help build Stream, a new conversational computer. As a machine learning engineer, you will develop a system which spans voice, memory, and agentic control. This role is perfect for someone cares deeply about real-world ML deployment and human-in-the-loop agentic interactions.
- Role:
- ML Engineer, Audio
- Job Type:
- FullTime
Company profile
sandbar
sandbar.comSandbar makes Stream, a private voice ring for capturing notes, chatting through ideas, and dictating across any app. The ring uses push to talk and release to send, with silent haptic feedback confirming each action, and it also lets users resume and control music while their phone stays in a pocket. Stream works as a conversational interface that gives live answers out loud, and it is available for pre-order.
