Full-Stack Voice AI Engineer, Conversational Agents for Healthcare (ElevenLabs, Python, Telephony)

DISCLAIMER! This is an individual engagement, read this first You will do this work yourself. No agencies, no teams, no consultancies, no subcontracting, and no passing the work to anyone else. If you are an agency, or you plan to route this through other people, please do not apply. We check for this throughout the engagement. If at any point we find that you misrepresented yourself, that you are an agency, or that you are handing the work to someone else, we will end the contract immediately and report the misrepresentation to Upwork. Upwork's own terms prohibit this, so please do not try to game it. JOB TITLE Voice AI Engineer, Conversational Agents for Healthcare (ElevenLabs, Python, Telephony) JOB DESCRIPTION We are Linear Health, an AI-native healthcare operations platform in the US. We sit above the electronic health record and automate the outbound work clinics do by hand today: referrals, prior authorization, scheduling, care gap closure, and patient outreach across voice, SMS, and email. Our customers are healthcare providers across the US. We are lean, profitable, and moving fast. We are hiring one voice AI engineer who lives and breathes conversational AI. This is not a chatbot role. It is real-time voice, over the phone, in a regulated domain, where a dropped detail or a bad interruption is a real problem for a real patient. WHAT YOU WOULD BUILD Voice AI agents on our enterprise ElevenLabs stack, both single-agent and multi-step workflow architectures Conversational design: state machines, turn-taking, interruption and barge-in handling, confirmation before any action commits to the record Prompt engineering for voice, tuned for latency, accuracy, and natural pacing, in English and Spanish Backend tools in Python and FastAPI that read and write patient data, appointments, and referrals in the EHR Telephony over Twilio PSTN and SIP, and the messy realities of real calls: noise, accents, silence, wrong numbers Escalation paths, quiet-hours rules, and safe handling of protected health information WHO WE ARE LOOKING FOR We have learned exactly what makes someone succeed here, so we screen for it on purpose: You pick things up fast. We explain a workflow once and you run with it. If you need it explained three times or need to be walked through every step, this is not the seat. You take full ownership. You carry work all the way to done without being chased, you keep your files and your work where they belong, and you close your own loops. Nobody should have to chase you for basics. You are proactive. You spot gaps, ask the right questions early, and flag problems before they grow, instead of going quiet or waiting to be told. You communicate. Strong, clear written and spoken English, because you will be on live calls, in customer meetings, and writing customer-facing messages. You take deadlines and reliability seriously. You do not need US healthcare experience. You do need to ramp fast and be comfortable in a regulated, high-stakes domain. MUST HAVE, AND BE READY TO PROVE Real production experience building voice AI agents, not just text chatbots. Tell us what you shipped and what it did in the real world. Strong grasp of conversational state design and interruption handling. If barge-in and turn-taking are new terms to you, this is not your role. Python and FastAPI for the backend tools behind the agent. Fluent, confident spoken and written English. Bonus: Spanish, telephony depth (SIP, WebRTC, PSTN), latency optimization, and any healthcare exposure. Compensation, published up front A flat monthly rate of 1,500 USD. On top of that we provide a Claude Max 20x license and an AI tooling allowance worth about 250 USD per month, because we build with AI in the loop every day and we want you fully equipped. Total value is around 1,750 USD per month. This rate is standardized for everyone joining at this level and grows as you prove yourself. If this number does not work for you, please do not apply, we would rather not waste your time or ours. THE ENGAGEMENT Long-term, full-time. The first 90 days are a paid trial while we both confirm the fit. We invest heavily in the people who clear it. You must overlap at least 6 hours every working day with US Central Time. For most of the world that means working into your evening or night. Only apply if that genuinely fits your life. The work is intense and the bar is high. We move fast. HOW WE WORK, AND WHO SHOULD APPLY We are a small, flat, fast team, and we hold a real bar. Apply only if this is the environment you want, or the one you are hungry to grow into: We live on Slack all day. You are online, reachable, and quick to respond during working hours. Daily standup is mandatory. A weekly team meeting is mandatory. We build spec-first with AI in the loop. You write clear specs, direct AI coding agents, and still understand and own every line that ships. You are customer-facing: customer calls, customer emails, customer presentations. If you want a quiet, heads-down, low-communication role, please do not apply. To confirm you read this in full, begin your proposal with the word "barge-in".