New: claims dashboard and smarter insurance verificationSee what's new
VoxHealth
← All open positionsEngineering

Senior AI Engineer

Design and ship the voice AI agents that answer, book, and follow up for healthcare practices, from LLM orchestration to real-time speech.

Remote (US) or San Francisco · Full-time

About the role

Our agents hold real phone conversations with patients and pet owners: booking, rescheduling, refill requests, insurance questions, and handoffs to staff, in many languages. Every call is a test of latency, accuracy, and judgment.

You will own the quality of those conversations, taking ideas from prototype to reliable, measurable production impact.

What you'll do

  • Improve how the agent follows instructions, calls tools, retrieves practice information, and keeps context across a call.
  • Pick and route models for each task based on latency, cost, and quality.
  • Build the evaluation system: simulated calls, regression suites, and failure analysis on real production calls.
  • Work on the real-time voice stack: speech recognition, text-to-speech, turn-taking, interruptions, and multilingual calls.
  • Ship guardrails and monitoring so agents behave safely in production and within HIPAA requirements.
  • Listen to calls and talk with front-desk staff to understand the workflows you are automating.

What success looks like

  1. First 30 daysYou have shipped a measurable improvement to a live agent and know our eval and call-review tools well.
  2. First 90 daysYou own a core part of the agent (for example, scheduling accuracy or latency) with a dashboard showing it getting better.
  3. First 6 monthsOur evaluation system catches regressions before customers do, and you are setting the technical direction for the agent.

What we're looking for

  • 5+ years of software engineering, with 2+ years building LLM-powered products in production.
  • Deep experience with prompt and tool design, retrieval, and evaluating non-deterministic systems.
  • Strong Python or TypeScript, and good judgment about when an experiment is worth running.
  • You measure before you claim, and you are comfortable reading through hundreds of transcripts to find the pattern.

Nice to have

  • Real-time voice: speech-to-text, text-to-speech, telephony (SIP/Twilio), or speech-to-speech models.
  • Fine-tuning or distilling models for latency and cost.
  • Experience with healthcare data or HIPAA-regulated systems.

Benefits

  • Competitive salary plus equity
  • Health, dental, and vision insurance
  • Remote-friendly with flexible hours
  • Unlimited PTO
  • Annual learning budget
  • The equipment you need to do your best work