Staff Software Engineer @ 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.

About the Role We’re looking for a strong generalist engineer to connect Stream to the software people already use. Stream is a self extension that listens, remembers, and helps you think through conversation, and the more of your world it can reach, the more useful it becomes.

Most software wasn’t built to work with an AI. Making it work means understanding how each application is built, finding the most reliable way in, and making that connection dependable for people who rely on it every day. You’ll own this work end to end, work directly with the founders, ship on the same real-time Python services the rest of Stream runs on, and help shape how others build on Stream.

Responsibilities

• Build integrations between Stream and the apps, services, and AI tools people already use • Design how tools plug into Stream’s agent: tool definitions, result shaping, confirmation, and context passing • Own the integration layer end to end: OAuth and token lifecycle, least-privilege scopes, webhooks, rate limits, retries, and real-world failure modes • Build and operate MCP servers and clients in production • Work out reliable integrations for software without clean APIs • Instrument every integration so breakage is caught before users notice • Carry a share of the core backend and its on-call rotation

Qualifications

• 5+ years building software, with strong Python and comfort owning a live, real-time service • Have shipped third-party integrations in production: OAuth 2/OIDC, token refresh and revocation, webhooks, API versioning, and vendor quirks • Understand software architecture well enough to work with applications you didn’t write • Have integrated tools into an LLM agent that real users depend on, and been accountable for actions completing • Hands-on MCP experience • Generalist range across backend, auth, client code, and agent logic; can take an open-ended problem to a shipped result without a spec • Product sense for software that acts on someone’s behalf, including when the user isn’t looking at a screen • Communicate precisely and make tradeoffs explicit