Apply for this role
Fill in the quick form below — it takes a minute. Go to the application form →
Careers · Engineering · Taipei
Voice Platform Engineer
A phone call is unforgiving. Half a second of added latency and a conversation feels broken; a dropped packet and a customer hears silence. We're building a voice platform where an AI agent can hold a natural conversation — and that only works if the media layer underneath is rock solid. That's the job.
About the role
Real-time, or it doesn't count.
You'll build the voice stack: placing and answering calls, running call-control flows and IVR, and moving audio between the caller, our speech services, and the AI agent — all in real time.
This is media-heavy engineering where milliseconds matter. You'll work on call control, streaming audio, and the interfaces that let a voice agent listen, think, and speak back without an awkward pause. You'll sit with the Taipei voice team and work closely with the AI engineers in San Francisco who build the agent runtime on top of what you deliver.
What you'll do
- Build and operate the call-control layer — placing, answering, routing, transferring, and recording calls — exposed through our Voice API.
- Handle real-time media: streaming audio to and from speech-to-text and text-to-speech, and into the agent loop, with tight latency budgets.
- Implement call-flow primitives (say, play, gather, transfer, connect-to-agent) and the events that report on a call's lifecycle.
- Keep quality high across networks and regions — jitter, packet loss, and codec choices are your problem to tame.
- Instrument everything, and carry a share of the on-call rotation for voice services.
What we're looking for
- 4+ years of backend or systems engineering, with time spent on real-time or media-heavy systems.
- Hands-on with voice or streaming technologies — SIP, RTP, WebRTC, or media servers like FreeSWITCH or Asterisk.
- Strong in a systems language (Go, Rust, C++, or similar) and comfortable reasoning about latency end to end.
- A debugging instinct for the messy real world: bad networks, flaky carriers, and timing bugs.
- Working proficiency in English.
Nice to have
- Experience integrating speech-to-text or text-to-speech at scale.
- Familiarity with telephony interconnects and carrier-grade voice.
- Exposure to conversational or voice-agent systems.
Ready to apply?
Send us a call you'd be proud of.
Résumé plus a couple of lines on the real-time system you've built or fixed. We read every one.
How we hire
Four steps, usually two to three weeks. See the process →
Apply
Submit your application
Fill in the form and add a link to your résumé or portfolio. A real person on our team reviews every application — usually within a few days.
✓ We reply to every applicant✓ No account or login required✓ Your details go straight to our hiring team