Position Summary
Are you ready to make a real impact on people's lives and be part of a rapidly-growing team? Natera is seeking a Software Engineer III to help design, develop, and maintain our Voice AI platform. This platform handles thousands of patient calls daily, providing automated test status, identity verification, billing support, and intelligent routing — directly improving patient access to their genetic testing results. Join us in our mission to change the way disease is managed, and be a part of a dedicated group of professionals who are passionate about making a difference.
The Software Engineer III – Voice AI is responsible for building and maintaining the real-time conversational AI systems that power Natera's automated patient call center. This role requires strong expertise in TypeScript and Node.js, hands-on experience with voice AI pipelines (STT, LLM orchestration, TTS), and familiarity with telephony systems, agentic architectures, and event-driven design. You should be comfortable working with real-time audio streaming, WebSocket protocols, and the unique latency and UX challenges of voice-based AI. You'll collaborate with cross-functional teams in a fast-paced environment to ship features that measurably improve call efficacy and patient satisfaction.
Primary Responsibilities
Lead
- Take ownership of assigned voice AI features and components, guiding them through the full software development lifecycle.
- Contribute to design discussions, code reviews, and best practice adoption within the Voice AI team.
- Drive technical decisions on voice pipeline optimization — VAD tuning, turn-taking, interruption handling, and latency management.
Manage
- Plan and prioritize tasks in an Agile environment, ensuring timely and high-quality delivery.
- Work with Product Managers and stakeholders to refine requirements and scope technical efforts for conversational AI features.
- Monitor voice platform health metrics (call efficacy, ASR accuracy, per-segment latency) and prioritize improvements based on data.
Nurture
- Mentor junior team members, sharing knowledge and best practices in voice AI architecture, TypeScript, and real-time systems.
- Encourage a culture of continuous learning and technical excellence through pair programming and design reviews.
- Document voice AI patterns, integration contracts, and operational runbooks for the team.
Collaborate
- Partner with Product Managers, QA, and clinical operations to gather requirements, validate conversational designs, and guide projects from inception to deployment.
- Coordinate with other engineering teams to integrate voice agents with internal services via authenticated APIs.
- Work with the analytics team to ensure voice metrics flow correctly through the data pipeline for reporting and optimization.
Effect Change
- Drive improvements in our multi-agent orchestration approach — tool calling patterns, agent handoff logic, and state management across conversation turns.
- Advocate for high-quality standards and automated testing strategies for conversational AI systems, including voice-specific test patterns (simulated calls, transcript validation, latency benchmarks).
- Identify and resolve voice-specific UX issues: ASR errors on medical terminology, silence detection tuning, barge-in recovery, and end-to-end response latency.
Qualifications
- 5+ years of overall software development experience, focusing on scalable backend services using Node.js and TypeScript.
- 1+ years of experience with voice AI, conversational AI, or real-time audio systems in production.
- Hands-on experience with agentic LLM architectures — tool calling, multi-agent orchestration, prompt engineering, and conversation state management.
- Familiarity with voice AI pipeline components: STT (Deepgram, Azure Speech, OpenAI Whisper), TTS (ElevenLabs, OpenAI, Cartesia), and LLM APIs (OpenAI Realtime API, Anthropic Claude).
- Experience with telephony systems — Twilio (media streams, SIP, IVR) or equivalent WebSocket-based audio streaming platforms.
- Understanding of voice-specific challenges: VAD configuration, turn-taking, interruption handling, latency budgets, and audio codec management (mulaw/PCM).
- Solid understanding of the software development lifecycle (SDLC), including build, configuration, release, and deployment.
- Knowledge of microservice architecture and distributed systems best practices.
- Proficiency with AWS services (ECS Fargate, Lambda, DynamoDB, S3, Kafka/MSK, API Gateway).
- Experience with event-driven architecture and message processing (e.g., Apache Kafka, SQS).
- Strong relational database skills (MySQL) and exposure to NoSQL databases (DynamoDB, Redis).
- Demonstrated teamwork skills and a collaborative mindset.
- Excellent communication and organizational skills.
- Experience with RAG architectures (AWS Bedrock, vector stores, embedding models) is a plus.
Knowledge, Skills, and Abilities
- Node.js/TypeScript: Proficiency with frameworks such as NestJS or Express; strong async/await and streaming patterns.
- Voice AI Pipeline: Understanding of end-to-end voice flow — telephony ingress, STT transcription, LLM processing, TTS synthesis, audio egress.
- Agentic Architecture: Experience with multi-agent systems, tool calling (OpenAI function calling, @openai/agents SDK or equivalent), agent handoffs, and conversation memory.
- Telephony: Twilio media streams, WebSocket audio streaming, SIP, IVR routing, call recording.
- LLM Integration: OpenAI Realtime API, Anthropic Claude, prompt engineering for conversational agents, structured outputs.
- Database Technologies: Strong SQL skills (MySQL) and exposure to NoSQL (DynamoDB, Redis/ElastiCache).
- AWS: ECS Fargate, Lambda, DynamoDB, S3, Kafka/MSK, API Gateway, Bedrock, CDK.
- Event Streaming: Apache Kafka for analytics pipelines; SQS for job queuing.
- Authentication: Okta JWT flows, OAuth2 client credentials, service-to-service auth patterns.
- Containerization: Docker, ECS task definitions, Fargate deployment.
- CI/CD: GitLab (or other pipelines) for continuous integration and deployment.
- Testing & QA: Familiarity with frameworks like Jest; experience testing conversational systems (transcript validation, simulated call flows).
- System Monitoring & Troubleshooting: Datadog APM, LLM Observability, or equivalent tools for latency tracking and voice pipeline health.
- Compliance: Awareness of HIPAA requirements for voice systems — PHI handling, zero-retention patterns, encrypted storage.