• * A short intro (video, voice, or call) is required! Details at the bottom.**
About us:
We're a healthcare SaaS startup building AI voice-calling agents that handle phone calls, plus a web app for insurance claims, status tracking, and denials. Our web app includes real-time voice/telephony and insurance/EDI APIs.
We're launching soon and have a big wait list and will be scaling fast as we've partnered with many industry insiders. We want a dependable engineer for a steady, ongoing part-time role, keeping the platform healthy day to day, with hours that grow as we add customers. Long-term relationship.
The role:
This is a monitoring + maintenance + light-build role. The product already exists, works, and runs on AWS. We need someone to keep it running, catch problems before customers do, fix small issues, and ship improvements over time. You'll work alongside AI dev tools (we build with Claude/Cursor), so we care less about raw wizardry and more about someone who understands WHY things work, notices subtle problems, and communicates clearly.
What matters most to us:
We need someone whose instinct is to make failures loud, the kind of engineer who, dropped into a system, immediately asks "why is there no alert on this?" In practice: monitoring and alerting, validating that data is actually good before we trust it, keeping per-minute and per-token costs from spiraling, and debugging quickly and calmly when something breaks. This is the heart of the role.
A very close second — real-time voice / telephony. Our core is AI voice agents (LiveKit / Twilio / SIP / WebRTC), and it's the trickiest, most expensive part of the stack. You don't need to have architected a voice engine from scratch, but you should be comfortable reasoning about and debugging one, keypad/DTMF tones, when the agent should speak, call transfers, endpointing on long holds.
What you'll do
• Make failures loud: build and maintain monitoring & alerting so we hear about a bad verification or a cost spike before a clinic does.
• Keep the AI voice-calling pipeline healthy (Python / LiveKit / Twilio / Deepgram / ElevenLabs) and debug it when a insurance line changes their phone system.
• Guard data quality: make sure a call that captured nothing never gets marked "done."
• Watch cost: per-minute calls + per-token AI, add caps, dedup, and spend alerts.
• Maintain the web app (Next.js / Postgres) and our AWS HIPAA stack (RDS, ECS/Fargate, Cognito, S3, CloudWatch). Small fixes and improvements as they come up.
Required
• Reliability instinct — you naturally reach for monitoring, alerting, validation, and cost guardrails; you debug production calmly and methodically.
• Solid full-stack: TypeScript / Next.js / React AND comfortable in Python.
• Real-time voice / telephony literacy (LiveKit, Twilio, SIP, WebRTC, or AI voice agents) — enough to debug and improve the pipeline.
• PostgreSQL, including row-level security (RLS) / multi-tenant data isolation.
• Hands-on AWS: RDS, ECS/Fargate (or App Runner), Cognito, S3, CloudWatch, IAM, Secrets Manager. Terraform / IaC a plus.
• Clear spoken English (we do regular video calls); reliable weekday availability; comfortable with AI dev tools.
Nice to have
• Observability tooling (Sentry, Datadog, CloudWatch dashboards, structured logging).
• Healthcare EDI / insurance-claims APIs (clearinghouses like Stedi, Availity, Change Healthcare).
• Healthcare / HIPAA / PHI-handling experience.
How to apply, two quick steps (generic proposals get skipped):
• Step 1: answer these 3 questions:*
1. Tell us about a time you added monitoring/alerting or caught a production problem before users did. What did you build, and how did it help?
2. Have you built or maintained a real-time voice or telephony system (LiveKit, Twilio, SIP/WebRTC, or AI voice agents)? What was it, and what was the hardest part to get right?
3. Tell us about a production bug you debugged in a system you didn't build, how did you find the root cause?
• Step 2 — send a 60–90 second intro (video OR voice note, your choice):* who you are, one relevant project, and why this role. Prefer not to record? A quick 10-minute intro call works too. (We just want to hear you before we set up an interview.)
Details. $18–30/hr depending on experience · ~15 hrs/week to start (about 2–3 hrs/day) · ongoing & long-term, hours grow as we add clinics · some overlap with US hours preferred.
