We're hiring a Voice AI Engineer to work on Parrot — our voice pipeline that powers AI communications across Ryft's portfolio companies. This is a hands-on, problem-solving role. You'll be improving and extending a real-time voice system that sits at the intersection of telephony, LLMs, and healthcare workflows.
The work
-> 50% voice pipeline: latency, audio quality, provider integrations, edge cases.
-> 50% the LLM side of the voice pipeline: context handling, response quality, model behaviour.
-> Building novel solutions to hard problems, not stitching together tutorials.
-> Node / GCP. That's the stack. It's boring on purpose.
What we're looking for
-> You've built something with AI voice recently — not just "I made a VAPI agent," but you understand what's happening underneath
-> You've worked with voice at a component level: STT, TTS, interruption handling, streaming, provider SDKs
-> You understand how LLMs actually work — not just prompting, but the mechanics of why they behave the way they do
-> You can look at a system, understand how and why it works, and make it better
-> You solve complex problems in simple ways. No overengineering, no shortcuts.
-> Senior or strong intermediate. You can work autonomously and have detailed technical discussions — not just executing on Linear issues.
Working at Ryft
-> Fun product experiments
-> Small team, moving fast
-> Work from whatever
-> Very few meetings
The people we look for
-> Chaotic good
-> Not afraid to fuck up
-> Strong craft focus
-> High agency
Application Process
-> Fill the application below
-> No resume required, just your social links
-> Show us something you've built with voice or LLMs
We'll review your bio/posts/website/github and reach out if I think there may be a fit.
