About Valene Health
Valene Health is building an AI-native clinical automation platform for healthcare organizations. Our agentic AI workforce helps providers and care teams reduce administrative burden, improve patient communication, and expand care capacity through intelligent automation.
Our platform supports AI voice agents, patient outreach, scheduling coordination, post-discharge follow-up, clinical documentation, care navigation, and other operational workflows. We work with health systems, physician groups, specialty practices, FQHCs, CCBHCs, MSOs, and risk-bearing organizations.
Role Overview
We are looking for an early-career applied AI engineer to help build the voice agents at the core of Valene Health’s platform. You will work on how agents listen, respond, use tools, extract information, and escalate concerns within defined clinical workflows. This is a product engineering position—not a pure research role—for someone who wants to turn modern LLM and speech technology into dependable, patient-facing experiences.
Key Responsibilities
- Build and improve the real-time voice pipeline, including speech-to-text, LLM orchestration, text-to-speech, and telephony integrations.
- Translate configurable call steps, instructions, capabilities, and escalation rules into reliable prompts, tool calls, and agent behavior.
- Use LLMs to extract structured clinical context from procedure and discharge documents.
- Generate concise call summaries, symptom reports, and escalation flags for clinical staff.
- Create test scenarios, evaluation datasets, and quality checks for call flow adherence, safety escalations, and extraction accuracy.
- Improve latency, reliability, and conversation quality across interruptions, silence, varying accents, and other real-world audio conditions.
- Collaborate with product, engineering, and clinical stakeholders to identify failure modes and improve agent guardrails.
Qualifications
- Solid Node.js and TypeScript skills, with experience building and debugging working applications.
- Practical experience with LLM APIs, including prompting, structured outputs, tool calling, and handling unexpected responses.
- Exposure to speech, audio, or real-time communication systems through work, coursework, or projects.
- Ability to write tests, work with Git and GitHub, and collaborate through pull requests.
- Thoughtful judgment about reliability and safety, including when an AI agent should defer, refuse, or escalate.
- Curiosity, adaptability, and comfort owning technical problems in an early-stage environment.
Preferred Experience
- Deepgram, ElevenLabs, or comparable speech and voice services.
- WebSockets, streaming audio, SIP, or telephony platforms such as Twilio and Telnyx.
- LLM evaluations, automated prompt testing, or LLM-as-judge approaches.
- PostgreSQL, APIs, and backend service development.
- Exposure to healthcare workflows, clinical documentation, or regulated technology.
What Success Looks Like
- Reliable voice workflows that follow instructions and escalate appropriately.
- Measurable improvements in call quality, latency, extraction accuracy, and agent behavior.
- Repeatable evaluation coverage for high-priority patient scenarios and edge cases.
- Clear collaboration with the broader team to turn field observations into product improvements.